Open AI Versus Open Ai? What can a trademark lawsuit tell us about the future of artificial intelligence and about what makes winners and losers in Silicon Valley? By Evan Ratliffe. It sounded like the setup to a joke, and for a while I thought it was a company called open AIS. Who's a company called open Ai? The case, formerly named open Ai Inc. Versus Open Artificial Intelligence, Inc. Showed up on the docket for the U. S. District Court in Northern California to zero fanfare in August of last year. Companies with near identical names entangled in a lawsuit usually means trademark infringement, more the makings of a one liner than a riveting story. Still, it did involve the open Ai, the generative AI kingpin and maker of chat gpt, backed by Microsoft to the tune of at the time thirteen billion dollars. I pulled some of the case documents and immediately found what seemed like the punchline. Open Artificial Intelligence, Inc. Aka the other open AI with a space was in fact just one guy named Guy. Guy Ravine, according to the suit, was a self styled Silicon Valley technologist who'd managed to grab the url open do ai back in twenty fifteen, before open ai no Space launched. When Sam Altman and Greg Brockman announced their venture to the world that December eleventh, they'd been forced to go with the less buzzy OpenAI dot com. Ravine then appeared to have moved beyond domain squatting, filing for a trademark on the name open ai with a space on the very evening Altman and Brockman made their announcement. Scanning the documents, I had to respect the hustle, Ravine asserted fantastically that he'd been working on an idea identical to Altman and Brockman. This wasn't the first time he'd had a world changing innovation yanked out from under him either. He professed to have invented the video sharing technology later made famous by Snapchat and TikTok. For someone making such sensational claims, however, Ravine had almost no online footprint. What little biographical information I did find lived on ancient looking websites, sometimes with the same low res photograph of a smiling man with the receding hairline. I suspected he might be a scammer or a kook, and open AI's lawyers seemed to agree. Their filings unsubtly mocked Ravine and implied a shakedown. One quoted an email Ravine sent to Altman in twenty twenty two, which noted that Elon Musk paid eleven million dollars for the Tesla domain and trademark in twenty seventeen. As we both know, OpenAI holds the potential to become larger than Tesla, so the ultimate value of the domain and the brand are substantial. Open ai asked the Federal Joe Judge Yvon Gonzales Rogers to bar Ravine from using the name until the lawsuit was resolved. It seemed that his quixotic argument about a parallel open artificial intelligence effort stood little to no chance up against some of the most formidable lawyers in the Bay Area, backed by open aiye's functionally infinite pockets. Sure Enough, in February, Gonzales Rogers issued a preliminary injunction against Ravine, forcing him to take down his website and delete all references to open ai. The evidence, she wrote, paints a troubling picture of defendant Ravine's representations. She strongly implied, in other words, that he was full of it. Ravine fired his attorneys that I figured was pretty much that. But then in April, a curious document appeared on the court docket. Ravine's new lawyers had filed a one hundred page countersuit against open ai, Altman and Brockman. In it, Ravine claimed that in twenty fifteen he'd actually been in discussions to raise one hundred million dollars for his open source artificial general intelligence or AGI project. He'd pitched open AI, sometimes calling it open dot ai. Two Silicon Valley luminaries, including Google's Larry Page, Meta Platform's chief AI scientist Jan LeCun, Google director of research Peter Norvig, Stripe Chief executive officer Patrick Collison, and former Apple executive Tom Kruber, the co creator of Siri. This wasn't a case of parallel invention, Ravine alleged, but of theft Altman and Brockman stole their recipe from him. The countersuit included a supposed statement from LeCun saying more people need to know about this, and from Gruber it is tragic that the hijacking of open AI from Guy Ravine may have historic consequences. Honestly, it sounded like madness one hundred million dollars historic consequence. I wondered if Ravine invented the quotes to back up his unhinged assertions. So I contacted Gruber, a highly respected voice in AI, and asked him if he was aware of the lawsuit. Of course, he said he'd given an official declaration. Wait, did he believe Ravine was for real? Gruber was unequivocal. He was a serious AI dude from the beginning, he said, I have my email records from it. I'm absolutely sure he was pitching me open AI at least six months before Altman showed up. Did that mean Ravine was really talking to those other luminaries, I mean I saw him pitch to Larry Page, Gruber said. In his telling, Ravine was hardly the grifter portrayed in the lawsuit. He was a righteous underdog. He's not a charismatic, you know, Sam Altman, Gruber said, But he's smart and he's honest. He'd gotten there first, trademarked his work, and now open AI was trying to erase him from the story. Grueber, who's gone on to advise other AI efforts since leaving Apple, even invested in a company Ravine founded. As for who was shaking down whom, Gruber reminded me that it was open Ai who sued Ravine, not the other way around. It's just not fair, That's all I'm saying. The joke I realized might be on me, so I decided to seek out Guy Ravine myself. The story I found was less about a simple trademark dispute and more about the struggle to turn ideas into reality, and about who and what determined Silicon Valley's winners and losers to open Ai. It may have seemed a simple case of a jealous troll appropriating its hard won success. We took legal action to stop the intentional use of open ai from confusing and misleading our users, a company spokesperson said, adding he also claims to have invented various successful tech companies in the past. But for Ravine, the case was about a vision, his vision for a powerful technology controlled not by companies but by all of humanity, and how its appropriation poisoned an industry from inception. The absurdity of this I cannot even describe it, Ravine says, his Israeli accent, catching me by surprise as we sit down across a table from each other. Knowing nothing about him, I'd assumed he was American. We're in the lush backyard of his rented one story house in Sunnyvale, modest by valley standards. When I'd pulled up, the open garage was cluttered with what appeared to be a dozen identical desk lamps. A disconnected washing machine sat where a car would go. Ravine doesn't own one, opting for ride share when he leaves home, which these days is rarely. He looks approximately like his one low res photo online, but more slight, younger than his forty three years In a black Nike baseball cap. He's wearing jeans and a loose fitting white Henley T shirt, the same outfit I'll find him wearing the next day. Seemingly having taken the start up uniform cliche to the extreme. He has stacks of jeans and white T shirts in the living room he appears to use as a closet. He'd actually contacted me before I reached out to him, inviting me to Sunnyvale, after hearing through the grapevine that I was looking into him. We would spend fourteen hours talking at his house over two days in May, and then a half dozen times afterward. He does not I can report come off as a cook, consumed, maybe intense. His thoughts tend to loop, returning to the same moments and phrases like their earworms. He's trying to exercise aloud. Facing the world's most powerful start up has also fueled a touch of paranoia. He asks me to put my cell phone on airplane mode before we begin talking. For reasons he keeps off the record. Mostly he seems baffled. They suit me first, he says, as we settle into the story, I'm a peaceful individual. I never intended to sue anybody. He'd spent decades keeping information about himself off the Internet, valuing humility and avoiding the distractions of social media. Now, he says, open AI is attempting to paint me as a fraud and a troll who came out of nowhere and basically tarnish and destroy my reputation. Periodically, as he tells the story, he turns to me with widened eyes and utters a kind of outraged mantra, what the f certain facts are not in dispute, the most significant being that Ravine registered the domain open ai in March twenty fifteen and owns a registered trademark on the words open ai, applied for that December. The questions at issue are why he did these things, what he planned to do with his intellectual property, and what he actually did. Ravine says the answer to the first question dates to two thousand and nine, when he saw a student presentation from a neural Network's Life at the Massachusetts Institute of Technology showing how advanced deep learning systems could identify visual scenes. Ravine, a largely self taught programmer, became convinced that deep learning would usher in a superhuman. AGI Inspired Ravine came up with an idea for an open collaboration platform he called wiccaneering wiki like Wikipedia, but for engineering. It would be a place where the smartest minds in AI and other technologies could work together, a collaboration that would eventually be joined by AI itself as it became more intelligent. We were very much like it's about human collaboration, but it's also about building the data inputs needed for machines to learn, says Kirk McMurray, an entrepreneur with a graduate degree in cognitive science who worked with Ravine on and off for a decade, including on Wiccaneering. Ravine connected with Eric Brynolfson, then the director of the MIT Initiative on the Digital Economy, who came an enthusiastic proponent of the platform. But over the next few years, Ravine says he and others saw that Google was cornering the market on deep learning, a realization cemented by the company's twenty fourteen acquisition of AI startup deep Mind. As a believer in the potential for AGI to surpass human intelligence, Ravine found this troubling. I thought, Okay, it's kind of dangerous that you're going to have this enormous concentration of power in the hands of a single corporation. He says. He considered starting his own company, but quickly abandoned the idea. First of all, I can't build AGI myself because even though I tried to dabble with neural networks, I realized that I don't have the prerequisite skills. He says. It's something that requires the best talent in the world to work on. That talent was concentrated at Google and was paid handsomely, so for any startup company to come out of left field and say we're going to compete with Google in artificial intelligence was considered ludicrous. There had to be another way. I spent three years thinking about this, he says, from twenty twelve to late twenty fourteen, and then finally a breakthrough. He would take the principles of wiccaneering, a nonprofit built around open research and collaboration, and build an organization to challenge Google in AI. You couldn't compete with them on money, he says, but you could compete with an ideology. In other words, promise researchers they'd be building AI in the open for the benefit of humanity instead of for profit, and you might appeal to the best ravine hit on a name that could instantly convey that ideology, Open AI. If any of this last reasoning sounds familiar, that's because it's identical to the argument Altman and Brockman would provide for starting open AI nearly a year later. In their telling, they came to these same conclusions in conversations with Elon Musk, Reid Hoffman, and others, particularly at an oft mentioned dinner at the Rosewood Hotel in Menlo Park. In the summer of twenty fifteen. Months earlier, however, Ravine had purchased the domain open dot ai for twenty five hundred dollars. He had meetings, dozens of them, with every prominent tech figure he could find. At the TED conference that March, he cornered Larry Page at Pages after party, where he also met Jan Tollen, co founder of Skype, and Gruber from Apple. Gruber was immediately taken with the notion Apple struggled to attract top researches in AI because its own researchers were not allowed to speak openly or publicly, he said in his declaration for the lawsuit. This made Guy's proposal of open AI as a way to attract researchers and accelerate progress while sharing the fruits with humanity a powerful idea which I personally supported by may. Ravine had emailed with Facebook now Metta's Lecoon and corresponded with Nick Bostrom, a prominent and controversial Oxford University professor who spent decades writing about AI safety. With the help of Bernyolfson, he wrangled a brief in person with Stripe CEO Collison, and then with Andrew Ing, chief scientist at Baidu and one of the world's leading AI experts. Ravine then followed up with Collison by email, laying out the principles of open Ai. Collison said through a spokesperson that he had no record of the meeting and declined to comment about the email. Hung didn't reply to a request for comment, according to emails. Brynnolfson also told Ravine he'd pitched Ravine's ideas to a wider set of luminaries, including TED conference head Chris Anderson, LinkedIn co founder Hoffman, and Musk. I think we've made progress with Reed and others, he wrote to Ravine in late March. Hoffman declined to comment. Ravine believed he needed one hundred million dollars in funding to make open ai work, but the commitments were slow in coming. Ravine was not a guy who'd even come close to raising that kind of money before, and in the email exchanges he showed me it's difficult to discern genuine interest from a polite brush off. Sebastian Thrune, a former Google vice president and co founder of Udacity, begged off an advisory role, saying he was over committed. Sounds exciting, wrote Norvig, the head of research at Google, but by then the proposal had morphed into a physical school to teach AI concepts. Ravine was hustling, and like most people applying the valley for money, he was twisting and turning as he waded through the polite condescension. Still, he remained optimistic. That September, he added a splash page to open dot ai, declaring that an announcement will be made soon. Then on December tenth, he emailed Bostrum an update, musing on a future where no one group has an algorithmic advantage in general AI. What he was launching, Ravine hoped would shift the critical mass of development to an open effort. He had a feeling that the coming weeks could have a big impact on the course of evolution in a I. He was right, but not in the way he expected. For years, Ravine had a habit of keeping a strange kind of to do list, not of bucket list items or daily tasks, he says, but of like the things that most people don't do because they are insane, things like running fifty plus mile ultra marathons or swimming in San Francisco Bay every day for a year without a wet suit, so it was that on the afternoon of December eleventh, he jogged down to the San Francisco waterfront, stripped to his shorts and dived in. Emerging from the frigid water a few minutes later, he grabbed his phone from his shoe and noticed a one line e mail from Chris Cadieu, an AI researcher and entrepreneur with whom he'd been discussing open ai seen this. It was followed by a link to a blog post at OpenAI dot com announcing open Ai No Space, founded by Altman Brockman and Iliya Sutzkver, an esteemed AI researcher lured from Google. Open ai described itself as a nonprofit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. The post read to Ravine like it was ripped from his own proposals. His first thought was who are these guys? And then what the f The post asserted that open ai had already garnered a billion dollars in funding commitments from the likes of Hoffman and musk I started thinking, Okay, well, maybe it's a good thing, Ravine says, because the whole idea here was to create it, make it open for humanity. The vision was the idea. It wasn't about making money or my own ego. But there was another voice in his head, he says, one that said they stole the whole idea, the whole co and the name. Now at this point you might be thinking, sure, guy Ravine came up with a similar idea to open Ai. Okay, he even named it before they did and pitched it to some of the same people. But open Ai launched. Open Ai didn't game over. This is the unwritten code of the start up world, writ large. The winners are those who build, and by that same code, there's no bigger loser than the also ran inventor claiming their idea was stolen and clamoring for a piece of someone else's success. Think of the Winklevoss twins, enshrined in tech lore as chumps after Facebook settled with them for eight figures. Or Kevin Halpern, who in twenty fifteen sued Uber Technologies for a billion dollars, claiming that founder Travis Khalanik had stolen his idea for a cell phone powered car service after Halpern discussed in a shared office space. The suit wound on for seven years before a California court finally threw it out and sanctioned Halpern. There's Patrick rash a British inventor who spent decades entangled in an ongoing eighteen billion dollar claim against Apple, alleging in various courts that it stole his ideas for iTunes. Rash As said he's received death threats from Apple fans. My kid was being bullied, he told The Guardian recently, with kids saying your dad says he invented this. He didn't. He stole it. Apple invented it. How one views these stories reflects something of our feelings about genius ownership and ruthlessness in an industry that thrives on appropriation. But for those on the losing end, who believe they're watching their ideas make others wealthy and influential, it must feel like being frozen in time. They remain poised at the moment when the idea had promise, believing but never knowing they could have made it too. Ravine grew up in a small villa in Israel, where his mother ran the family's successful construction business. His early character fit the tech founder archetype, smart, precocious. Given a PC at age five, Guy was like my chat GPT, his brother, Saggie Reuven says, Ravine changed his last name after arriving in the US, sick of having to spell it for people. If I had a question, I would go to my mom and my mom would say, I don't know, go ask Guy. Ravine got into programming and created an online poker game that nearly got him kicked out of school. At eighteen, he managed to get himself excused from military service and moved to England. I didn't believe in what the Israeli government was doing, and also I thought it wouldn't be a good use of my time, is all he'll say about it. While taking classes at the University of Warwick, he made what he says was his first foray into AI, a chat bot capable of looking up answers on the Internet. It wasn't great, he says, and I thought it would take a long time to make it great, but it was good enough to get a small amount of funding and contacts to make his way to MIT, he audited classes and joined the startup scene, posting meetings for a tech incubator called MIT Playground on campus flyers and online message boards. A half dozen kids showed up, including future Cora CEO Adam Dangelo, then a California Institute of Technology student who was visiting the Boston area, and future collaborator McMurray, then a grad student. The group was fascinated with the ways fledgling social networks like Friendster and MySpace were struggling with speed as they scaled in size. Ravine had a company called I Need a Bit like if Task Grabit were also a social network that he says Dangelo also worked on. Guy had a sort of solution, an algorithm that dealt with the scaling issues, says McMurray. In Ravine's telling, he coded up the algorithm and sent it to Dangelo to implement, and then nothing. He disappears, never heard back from him. Ravine says, where does he emerge next? Co founder and CTO of Facebook responsible for scaling? What the f walking Ravine through his career is essentially a procession of these wtfs in which Ravine hatches ideas that later become major technology successes. In two thousand and two, it was a company that would let people amplify and share their Wi Fi connections that inspired what later became a billion dollar unicorn, he says. In two thousand and three, it was Stayway, a social network Ravine built that would allow people to rent out their spare rooms. He shows me a snapshot of the company's website, a map of the world covered in dots for shared lodging. It's going to be a multi billion dollar company, he remembers telling colleagues people were looking at me and scratching their heads. Airbnb made its debut five years later. In two thousand and seven, Ravine became convinced that electric vehicles were the future and designed his own single occupant electric car. He mixed in the ev community with the likes of early Tesla founders Martin Eberhart and Mark Tarpining, and wrangled contracts to build a prototype. The market collapse of two thousand and eight ended the dream of funding it. He did remain in touch with Eberhard and Tarpining, he says, and in twenty twelve I tried to convince them to invest a million dollars in bitcoin when bitcoin was ten dollars apiece. When I express some skepticism about this, Ravine pulls up the emails of him doing so. Eberhart and Tarpening didn't respond you requests for comment. He then shows me later ones after the pair failed to invest, in which he's ribbing them for missing out. Ravine himself did invest and eventually made a small fortune. Another example, he says, of seeing a future that others can't. It's not that Ravine always believes ideas were stolen from him. Sometimes he says he was just too early. Other times he couldn't line up the funding. Occasionally, though, there's some thread of connection that, at least to him, hints at inspiration, if not appropriation. Perhaps the singular example of Ravine's Zelig like technology career was his video sharing idea, the claim that had convinced me he might be a kook. In twenty eleven, he and McMurray were working on a dating app when they discovered that the friends they'd enlisted to test it were obsessed with one feature, the ability to share quick smartphone videos to your social network, as well as swiped through your friend's videos, which played instantly. They built a company around it called We. It aimed to capitalize on, as their funding deck put it, the massive opportunity of creating a purely video communication platform that is self contained. Ravine showed me old videos from WI, and I have to say, other than the fact that its users shot horizontally, it does look like Snapchat video before Snapchat video, or TikTok before TikTok. But among the reasons video had yet to take off was the cost of delivering it. Like you couldn't grow to one hundred thousand users and pay your bills and worry about making money later. McMurray says, you needed to be a high growth venture funded startup. When they scoured the valley for capital at firms such as Founder's Fund, though investors demanded tens of thousands of users, the same users they needed the investment to afford. Finally, according to Ravine, light Speed venture partners seemed ready to put up one million dollars. He says, one partner bulked, declaring that video sharing would never work, and the funding never materialized. As it happened, that partner was on the board of Snapchat now Snap Inc. Which added swipeable video sharing to its disappearing photo service within the year, TikTok emerged out of China a few years later. Neither company responded to inquiries, but facts which to Ravine feel suspicious even to Farius, to others simply amount to coincidence and a typical Valley funding rebuff. The light Speed partner reveal Vi says was ready to fund him, didn't respond to requests for comment. The one who Ravine says nixed the deal says it's just unequivocally not true that he said video wouldn't work. He doesn't remember Ravine, but says he invested in another social video platform at that same time. At any given moment, the world recognizes that a certain trend is coming, and there are multiple different inventions that emerge, Ravine says. He maintains that his implementation took off with those later companies because it was the most immersive and more accessible. McMurray moved on to other things, while Ravine obtained a patent for some of the technology and pivoted to a business to business video company called video dot Io, which was a bad idea, he says, but that's what I could get funding for. He believes Snapchat and TikTok are likely infringing on the patent, but says he hasn't had the stomach for legal action. Still, in the lawsuit with open Ai, his filings refer to him as having invented the foundation of modern mobile video, a field that was swarming with activity during that time. It's the kind of claim that can make Ravine seem well out over his skis video dot Io, which allows companies to make their own TikTok like apps, proved successful enough for Ravine to make comfortable living on top of his nest egg of prescient crypto investments. But there's comfortable and then there's what could have been. People who know me, my friends, they've always asked, how come you're not a billionaire, he says. They've been asking this for twenty years. They saw me coming up with all these things that people use today, and they're either stolen by other people or we couldn't get the funding for it. I don't know what to make of it. I could be depressed about it, or I could just live my life. We'll be right back with open Ai versus open Ai. Welcome back to open Ai versus open Ai. The day in December twenty fifteen, when open ai was announced, Ravine ran home from the beach. He took the first of several crucial steps that would set in motion the situation he finds himself in today. First, he went on the US Patent and Trademark Office website and filed for the trademark on open ai with a space. He did this without consulting a lawyer and lacking any real trademark expertise. And then at nine twelve p m. He sent an email to Altman and Sutzkever, two of open AI's three co founders. We've been working on an initiative called open dot ai to build a collective engineering platform to enable researchers from around the world to collectively engineer deep learning algorithms, he wrote. The initiative has the same goals as yours, which are to accelerate the arrival of general AI through an open effort. He mentioned some of the people he'd spoken to, Bostrom and Peter Norvig, with whom he'd discussed the AI school idea. It would make sense for us to meet and see if we could team up, he wrote, because we are working toward the same goals. This is bigger than us, and I think it could benefit from working together. Altman copied in Brockman, who suggested they meet at the Tesla offices in Palo Alto. When he arrived the next week, Ravine says, Brockman was immediately dismissive of any collaboration, saying Ravine should abandon the idea and sell them open dot ai. With open Ai already launched and funded to the tune of a billion dollars, what was the point in him continuing? Ravine says. Brockman emphasized that open Ai would be working for humanity's benefit, but to Ravine, he was acting like they just launched a hot for profit startup. The lack of empathy here is absolutely astonishing, he recalls, thinking, I don't know what I said, something like I'm not interested in selling, and then I walked out. Open Ai declined to make Altman or Brockman available for comment. Sutzkever didn't respond to request for comment. Feeling defeated, Ravine turned back to wiccaneering and his day job at video dot Io. I got into a depression over this for two years, he says, I thought, what is my purpose? Then still he continued to pursue the open ai trademark application. In twenty sixteen, he got a notice from the USPTO that it was being provisionally rejected, saying the screenshots of open dot ai he'd submitted failed to show it being used in commerce, a trademark requirement. Ravine submitted more screen shots, which he claimed showed an active discussion group on the site in twenty fifteen. Later in court, his lawyers would admit that the screen shots were from twenty sixteen, saying he'd lacked the records and had tried to recreate the earlier state of the site. The USPTO accepted them, and the trademark was granted in twenty seventeen, though not on the principal register but the supplemental one, because the name open ai was descriptive rather than distinct. Even with the trademark registration in hand, Ravine seemed to let the site languish. In another step that would prove fateful, he began having open dot ai redirect or forward to OpenAI dot com OpenAI's site. He left it that way, according to records submitted by OpenAI's lawyers, for four years. Ravine now says he never stopped working on public facing AI experiments, which he says were hosted at subdomains of open dot ai that never redirected, but looking back, it's easy to conclude he'd seeded the field to Altman and company. Then, in January twenty twenty two, OpenAI applied for its own trademark. Suddenly, Altman emailed Ravine for the first time since twenty fifteen, Following up on past conversations. He noted that open dot ai was forwarding to OpenAI dot com, then asked if Ravine would be open to us acquiring the open dot ai domain and related IP rights from you. This was the infamous Tesla email cited by open Ay's lawyers in their suit against Ravine, in which Ravine noted that Musk had paid eleven million dollars for the Tesla site and trademark, but open AI's complaint had conveniently excluded the rest of Ravine's response. The issue is that if you offered me a sum, I have no use for the money, he wrote to Altman. As an individual, I'm already well off. Ravine suggested that open ai donate money to a non profit academic AI research initiative he planned to organize. If it did, he'd hand over the domain and trademark. When Altman asked for specifics, Ravine replied that he was busy working to get my Russian team and their families out in time, in an operation that turned into a James Bond plot that keeps getting more insane. The Ukraine War had just commenced, and Ravine tells me he eventually managed to get his video dot io programmers out on the last flight out of Russia. Basically, sorry to hear, great that you are doing that. Altman replied that was their final correspondence. He could have easily responded and said, O K, how much do you want me to donate? Ravine says, now, we would have had a conversation and he would have donated probably a few million dollars at the end. That September, open ai created a sensation by fully releasing the latest version of its image generator doll E. Then, two months later, for reasons that are a central dispute in the lawsuit, Ravine re launched open dot ai with a new tagline, imagine if the best AI models were open and free. On the site, he embedded a version of an open source image generator stable diffusion. By December, it had one hundred and seventy thousand users, according to Ravine's court filings. The question was whose users were they. Ravine wrote a letter of protest to the U s p t O contesting open AI's trademark application on the basis that it would be confused with his. Then he started making plans to move to France and live in a room quote cabin away from it all for a few years. But Ravine had poked the bear, and as he packed up his house on August eleventh, twenty twenty three, he opened an email from a lawyer at the firm Quinn, Immanuel, Erkhart and Sullivan informing him that open ai was suing him in federal court over the domain and trademark. Bob Feldman, a partner at the firm who's litigating the case, declined to comment. I'm like what the f Ravine recalls Altman. He says could have had it for free, or at least the cost of a donation. Instead, he decided to donate millions of dollars to literally the most feared law firm in the world to sue me again and again. In our conversations, he returns to that phrase, the most feared law firm in the world. Finally, I ask him how he knows this. He turns his laptop toward me and pulls up the email. The signature reads Quinn, Immanuel, Erkhart and Sullivan Most feared law firm in the world. It's hard not to see a hint of irony in open AI's claims. This is a company facing a raft of lawsuits from publishers and writers accusing it of stealing copyrighted works to train its models. Open Ai has said its use of copyrighted material for training data is covered under fair use doctrine. Here it is in federal court accusing a single person of trying to profit off its identity and years of hard work. Open AI's initial complaint alleged that Ravine had infringed on open AI's trademark, a trademark the company had filed for but not received, in part because of Ravine's existing registration. Ravine, in the lawsuits telling was the latecomer, confusing the millions of users of open AI's products into mistakenly believing that open ai had some connection to open Ai. Defendant's motives are transparent. It alleged, namely to profit from their misappropriation of open AI's established name, leverage for themselves open AI's goodwill, or fraudulently divert public interest in and demand for open AI's products. At the heart of the case are questions raised by open Ai about who established the name open ai space or No in the market first, and whether Ravine's trademark registration is even valid, calling on expert witnesses, The company accused him of fabricating the USPTO evidence that he was using the mark in commerce. Ravine counters that he filed the application without a lawyer, misunderstood the USPTO's requests for samples of the site, and attempted to recreate faithfully how it looked when he filed. Judge Gonzales Rogers wasn't buying it. Her injunction barring Ravine from using open ai during the case seemed like a bad sign for Ravine's chances, at least in her court. Ravine, in his countersuit, accused open Ai of a different type of deception, not of the USPTO, but of him and the public. Despite the contention in its twenty fifteen blog post that Musk and the others had already pledged a billion dollars. According to its own financials, open ai raised just under fourteen million dollars in its first year. The billion dollar claim, Ravine said, tricked him into abandoning his own venture. The countersuit was stuffed with other claims, too, ranging from the trivial to the outlandish. At one point, Ravine's filing implied without evidence that D'Angelo, the former chief technology officer at Facebook now CEO of Cora and open ai board member whom Ravine had known in the early two thousands, had used old credentials to wipe evidence off ravine servers. A subsequent refiling removed the accusation. De Angelo didn't respond to requests for comment sent to Cora's press office, and open Ai declined to make him available for comment. As for how exactly open AI's founders had poached his idea, Ravine's claims were purely circumstantial. In the spring of twenty fifteen, two months after Ravine met briefly with Stripe's Collison, Brockman, announced he was leaving Stripe as CTO. In a blog post, Brockman noted that he sought advice at the time from both Collison and Altman. My goal was to figure out what deep learning was, he wrote, Whether Collison or some other luminary mentioned Ravine's ideas to Brockman or Altman seems difficult to establish, even if the suit makes it through discovery. Collison declined to comment, but Ravine's countersuit also rested on another argument that open ai has no right to the trademark because it isn't in fact open at all. Ravine's lawyer cited, among other events, the firing and re hiring of Altman last November, followed by the purge of open AI's board and the departure of its safety team. Open ai has undergone a remarkable transformation, Ravine's claim stated, from an open nonprofit organization with a figuciary duty to humanity into a closed for profit entity that appears to be under the de facto personal control of a single individual, Sam Alton. It's humanity's asset, Ravine tells me. It's not his asset. In Ravine's telling, this transformation from open to closed isn't the result of some considered change as open ai evolved, It's the natural product of the company's original sin Trying to build an idea they didn't really believe in that wasn't theirs to begin with. The countersuit went so far as to ask the court to force Altman to change its deceptive and misleading name to closed AI or a different, more appropriate name. This, as it happens, is the nearly identical contention of Musk in a federal lawsuit filed on August fifth, accusing open Ai, Altman, and Brockman of deceiving him into giving forty four million dollars to a non profit that isn't Once OpenAI Incs Technology approached Transformative AGI, Musk's suit contends Altman flipped the narrative and proceeded to cash in. In partnership with Microsoft, Altman established an opaque web of four per off at open ai affiliates, engaged in rampant self dealing, seized open ai inks board, and systematically drained the nonprofit of its valuable technology and personnel. Musk, a professed believer in the potential for superintelligent AI to doom humans, is no disinterested investor, having introduced his own chat GPT competitor grock He also has long claimed he came up with the name open Ai. In a strange twist, Musk's case was randomly assigned to Gonzales Rogers, the same federal judge as open Ai versus OpenAI. Musk's attorney didn't respond to an inquiry. In early October, open Ai raised six point six billion dollars at a one hundred and fifty seven billion dollar valuation and is reportedly discussing completing a for profit transformation and offering an equity stake to Altman. In court, open Ai moved to dismiss Musk's suit, calling it the latest move in Elon Musk's increasingly blusterous campaign to harass open ai for his own competitive advantage. But when Musk had filed his similar lawsuit in state court this year later withdrawn, OpenAI published a blog post purporting to show that musk knew of the company's intentions. Included was a private email from Sutzkiver to Musk in January twenty sixteen, suggesting that the open in open ai was temporary, meaning only that everyone should benefit from the fruits of ai A pledge to share the company's research openly, it continued, was merely the right strategy in the short and possibly medium term for recruitment purposes. It was never about being open, Ravine argues, noting that the email dates to just weeks after Brockmann had emphasized that open ai would be of public good. They come to me in person and make this representation, he says, But they already had these plans back then. This is how insane. This is insanity. And then there's the law, and in trademark law, says Genie Fromer, a law professor specializing in intellectual property at New York University, it's all about use in commerce. That's the whole ballgame. Unless Ravine can prove he established what's called secondary meaning for open ai, some amount of real users associating it with artificial intelligence products before open Ai did, it's unlikely to matter how open ai came about or how it's behaved, since even Ravine's supplemental registration amounts only to a kind of public placeholder. If you establish secondary meaning, Fromer says, you have the rights. In late September, Judge Gonzales Rogers again sided with open Ai, granting a motion to dismiss the bulk of Ravine's countersuit. She ruled that Ravine's lawyers hadn't established that open AI's for profit transformation was legally relevant. Ravine says he plans to refile an amended version. He's already appealing the judges a ration regional preliminary injunction against him. One thing is certain. It'll take months or years of legal rankling and likely millions more dollars for the case to see trial if Ravine and open ai settle. Instead, the questions the case raises about open Aye's mission and whether it was tainted from the beginning or abandoned along the way, could quietly disappear. The tech industry, despite its veneration of risk, isn't known for its courage in speaking truth to money, Ravine says. When he originally tried to hire lawyers to defend against open AI's suit, nearly thirty Bay Area firms turned him down, preserving their ability to work with open ai and Microsoft. He eventually found all his attorneys, including a now third firm hired weeks ago in Los Angeles, and except for Gruber, the people who heard Ravine's vision back in twenty fifteen have stayed silent when the lawsuit was filed, Ravine says, memories suddenly grew hazy and once enthusiasm stick boosters stopped returning his emails and mine. Brynnolfson, the man who helped shop Ravine's idea to the valley's elite, never responded to me. LeCun sent me to Metta's public relations team, who declined an interview on his behalf. Sorry, I didn't have any recollection of conversations like that, Norvig, replied when I emailed to ask about Ravine. When I noted that Ravine had shown me emails of their correspondence, he followed up, I do have a memory of meeting with someone on these topics. It must have been guy Ravine. I remember telling him to focus on what he wanted to achieve and to let the domain name or copyright or whatever on open Ai go. Bostrom, who'd once asked Ravine to read a draft of a paper and give notes, remembered him now only as a guy he ran into at a couple different conferences. He says, I don't recall much about what we were talking about, beyond a vague memory that he was telling me about some AI or AI compute related project or projects that he was pursuing or thinking about pursuing. After all, the events happened a decade ago, and even if some people do remember things Ravine's way, there's little advantage for them in playing even a bit part in a lawsuit involving a tech juggernaut backed by the most feared law firm on the planet. In one sense, all the parties in both lawsuits are united by a single belief that Obenai is building the most powerful technology in human history, an AGI that will change society forever. Both Musk and Ravine are nominally engaged in a fight to stop Altman from being the one who controls it, but that's not really something the court can do for them. Even if they win. Musk might get his money back, but too late to seriously wound his AI competition, and even if Ravine's version of events is vindicated, he's likely to remain unwelcome in Altman's AI future. The lawsuit is now a year in with a trial date continuously receding over the horizon. It's taken over his life, Ravine says, forcing him to lose focus on his own company, Video dot Io, which he'd been on the verge of selling. He stopped exercising and abandoned his insane activities lists. He used to joke around a lot. His brother, Saggi says, he has like this fun, sarcastic humor, and now it's like turned off, like hey, where's the fun guy. It's affecting him. Ravine spends much of his time alone in his house, helping lawyers prepare filings and answering discovery requests. It all amounts to a daily retreading of his life's work, searching for ways to sum it all up that'll prove he's not a fraud and a troll. It makes my heart kind of heavy. Saggie says, It's like we have a saying in Hebrew, how can you prove you don't have a sister? So is Ravine the little guy with pioneering innovations screwed out of history by more connected and ruthless operators, or the sore loos dreaming up ideas that he doesn't have the wherewithal to develop. By now, you've probably made up your own mind. But I wonder too if Ravine doesn't represent something else, a stand in for the rest of us facing a future that threatens to run us over, enriching an industry with no time for the thwarted dreamers and never made it. Build your own, they say, or get out of the way.

Open AI Versus Open Ai? What can a trademark lawsuit tell us about the future of artificial intelligence and about what makes winners and losers in Silicon Valley? By Evan Ratliffe. It sounded like the setup to a joke, and for a while I thought it was a company called open AIS. Who's a company called open Ai? The case, formerly named open Ai Inc. Versus Open Artificial Intelligence, Inc. Showed up on the docket for the U. S. District Court in Northern California to zero fanfare in August of last year. Companies with near identical names entangled in a lawsuit usually means trademark infringement, more the makings of a one liner than a riveting story. Still, it did involve the open Ai, the generative AI kingpin and maker of chat gpt, backed by Microsoft to the tune of at the time thirteen billion dollars. I pulled some of the case documents and immediately found what seemed like the punchline. Open Artificial Intelligence, Inc. Aka the other open AI with a space was in fact just one guy named Guy. Guy Ravine, according to the suit, was a self styled Silicon Valley technologist who'd managed to grab the url open do ai back in twenty fifteen, before open ai no Space launched. When Sam Altman and Greg Brockman announced their venture to the world that December eleventh, they'd been forced to go with the less buzzy OpenAI dot com. Ravine then appeared to have moved beyond domain squatting, filing for a trademark on the name open ai with a space on the very evening Altman and Brockman made their announcement. Scanning the documents, I had to respect the hustle, Ravine asserted fantastically that he'd been working on an idea identical to Altman and Brockman. This wasn't the first time he'd had a world changing innovation yanked out from under him either. He professed to have invented the video sharing technology later made famous by Snapchat and TikTok. For someone making such sensational claims, however, Ravine had almost no online footprint. What little biographical information I did find lived on ancient looking websites, sometimes with the same low res photograph of a smiling man with the receding hairline. I suspected he might be a scammer or a kook, and open AI's lawyers seemed to agree. Their filings unsubtly mocked Ravine and implied a shakedown. One quoted an email Ravine sent to Altman in twenty twenty two, which noted that Elon Musk paid eleven million dollars for the Tesla domain and trademark in twenty seventeen. As we both know, OpenAI holds the potential to become larger than Tesla, so the ultimate value of the domain and the brand are substantial. Open ai asked the Federal Joe Judge Yvon Gonzales Rogers to bar Ravine from using the name until the lawsuit was resolved. It seemed that his quixotic argument about a parallel open artificial intelligence effort stood little to no chance up against some of the most formidable lawyers in the Bay Area, backed by open aiye's functionally infinite pockets. Sure Enough, in February, Gonzales Rogers issued a preliminary injunction against Ravine, forcing him to take down his website and delete all references to open ai. The evidence, she wrote, paints a troubling picture of defendant Ravine's representations. She strongly implied, in other words, that he was full of it. Ravine fired his attorneys that I figured was pretty much that. But then in April, a curious document appeared on the court docket. Ravine's new lawyers had filed a one hundred page countersuit against open ai, Altman and Brockman. In it, Ravine claimed that in twenty fifteen he'd actually been in discussions to raise one hundred million dollars for his open source artificial general intelligence or AGI project. He'd pitched open AI, sometimes calling it open dot ai. Two Silicon Valley luminaries, including Google's Larry Page, Meta Platform's chief AI scientist Jan LeCun, Google director of research Peter Norvig, Stripe Chief executive officer Patrick Collison, and former Apple executive Tom Kruber, the co creator of Siri. This wasn't a case of parallel invention, Ravine alleged, but of theft Altman and Brockman stole their recipe from him. The countersuit included a supposed statement from LeCun saying more people need to know about this, and from Gruber it is tragic that the hijacking of open AI from Guy Ravine may have historic consequences. Honestly, it sounded like madness one hundred million dollars historic consequence. I wondered if Ravine invented the quotes to back up his unhinged assertions. So I contacted Gruber, a highly respected voice in AI, and asked him if he was aware of the lawsuit. Of course, he said he'd given an official declaration. Wait, did he believe Ravine was for real? Gruber was unequivocal. He was a serious AI dude from the beginning, he said, I have my email records from it. I'm absolutely sure he was pitching me open AI at least six months before Altman showed up. Did that mean Ravine was really talking to those other luminaries, I mean I saw him pitch to Larry Page, Gruber said. In his telling, Ravine was hardly the grifter portrayed in the lawsuit. He was a righteous underdog. He's not a charismatic, you know, Sam Altman, Gruber said, But he's smart and he's honest. He'd gotten there first, trademarked his work, and now open AI was trying to erase him from the story. Grueber, who's gone on to advise other AI efforts since leaving Apple, even invested in a company Ravine founded. As for who was shaking down whom, Gruber reminded me that it was open Ai who sued Ravine, not the other way around. It's just not fair, That's all I'm saying. The joke I realized might be on me, so I decided to seek out Guy Ravine myself. The story I found was less about a simple trademark dispute and more about the struggle to turn ideas into reality, and about who and what determined Silicon Valley's winners and losers to open Ai. It may have seemed a simple case of a jealous troll appropriating its hard won success. We took legal action to stop the intentional use of open ai from confusing and misleading our users, a company spokesperson said, adding he also claims to have invented various successful tech companies in the past. But for Ravine, the case was about a vision, his vision for a powerful technology controlled not by companies but by all of humanity, and how its appropriation poisoned an industry from inception. The absurdity of this I cannot even describe it, Ravine says, his Israeli accent, catching me by surprise as we sit down across a table from each other. Knowing nothing about him, I'd assumed he was American. We're in the lush backyard of his rented one story house in Sunnyvale, modest by valley standards. When I'd pulled up, the open garage was cluttered with what appeared to be a dozen identical desk lamps. A disconnected washing machine sat where a car would go. Ravine doesn't own one, opting for ride share when he leaves home, which these days is rarely. He looks approximately like his one low res photo online, but more slight, younger than his forty three years In a black Nike baseball cap. He's wearing jeans and a loose fitting white Henley T shirt, the same outfit I'll find him wearing the next day. Seemingly having taken the start up uniform cliche to the extreme. He has stacks of jeans and white T shirts in the living room he appears to use as a closet. He'd actually contacted me before I reached out to him, inviting me to Sunnyvale, after hearing through the grapevine that I was looking into him. We would spend fourteen hours talking at his house over two days in May, and then a half dozen times afterward. He does not I can report come off as a cook, consumed, maybe intense. His thoughts tend to loop, returning to the same moments and phrases like their earworms. He's trying to exercise aloud. Facing the world's most powerful start up has also fueled a touch of paranoia. He asks me to put my cell phone on airplane mode before we begin talking. For reasons he keeps off the record. Mostly he seems baffled. They suit me first, he says, as we settle into the story, I'm a peaceful individual. I never intended to sue anybody. He'd spent decades keeping information about himself off the Internet, valuing humility and avoiding the distractions of social media. Now, he says, open AI is attempting to paint me as a fraud and a troll who came out of nowhere and basically tarnish and destroy my reputation. Periodically, as he tells the story, he turns to me with widened eyes and utters a kind of outraged mantra, what the f certain facts are not in dispute, the most significant being that Ravine registered the domain open ai in March twenty fifteen and owns a registered trademark on the words open ai, applied for that December. The questions at issue are why he did these things, what he planned to do with his intellectual property, and what he actually did. Ravine says the answer to the first question dates to two thousand and nine, when he saw a student presentation from a neural Network's Life at the Massachusetts Institute of Technology showing how advanced deep learning systems could identify visual scenes. Ravine, a largely self taught programmer, became convinced that deep learning would usher in a superhuman. AGI Inspired Ravine came up with an idea for an open collaboration platform he called wiccaneering wiki like Wikipedia, but for engineering. It would be a place where the smartest minds in AI and other technologies could work together, a collaboration that would eventually be joined by AI itself as it became more intelligent. We were very much like it's about human collaboration, but it's also about building the data inputs needed for machines to learn, says Kirk McMurray, an entrepreneur with a graduate degree in cognitive science who worked with Ravine on and off for a decade, including on Wiccaneering. Ravine connected with Eric Brynolfson, then the director of the MIT Initiative on the Digital Economy, who came an enthusiastic proponent of the platform. But over the next few years, Ravine says he and others saw that Google was cornering the market on deep learning, a realization cemented by the company's twenty fourteen acquisition of AI startup deep Mind. As a believer in the potential for AGI to surpass human intelligence, Ravine found this troubling. I thought, Okay, it's kind of dangerous that you're going to have this enormous concentration of power in the hands of a single corporation. He says. He considered starting his own company, but quickly abandoned the idea. First of all, I can't build AGI myself because even though I tried to dabble with neural networks, I realized that I don't have the prerequisite skills. He says. It's something that requires the best talent in the world to work on. That talent was concentrated at Google and was paid handsomely, so for any startup company to come out of left field and say we're going to compete with Google in artificial intelligence was considered ludicrous. There had to be another way. I spent three years thinking about this, he says, from twenty twelve to late twenty fourteen, and then finally a breakthrough. He would take the principles of wiccaneering, a nonprofit built around open research and collaboration, and build an organization to challenge Google in AI. You couldn't compete with them on money, he says, but you could compete with an ideology. In other words, promise researchers they'd be building AI in the open for the benefit of humanity instead of for profit, and you might appeal to the best ravine hit on a name that could instantly convey that ideology, Open AI. If any of this last reasoning sounds familiar, that's because it's identical to the argument Altman and Brockman would provide for starting open AI nearly a year later. In their telling, they came to these same conclusions in conversations with Elon Musk, Reid Hoffman, and others, particularly at an oft mentioned dinner at the Rosewood Hotel in Menlo Park. In the summer of twenty fifteen. Months earlier, however, Ravine had purchased the domain open dot ai for twenty five hundred dollars. He had meetings, dozens of them, with every prominent tech figure he could find. At the TED conference that March, he cornered Larry Page at Pages after party, where he also met Jan Tollen, co founder of Skype, and Gruber from Apple. Gruber was immediately taken with the notion Apple struggled to attract top researches in AI because its own researchers were not allowed to speak openly or publicly, he said in his declaration for the lawsuit. This made Guy's proposal of open AI as a way to attract researchers and accelerate progress while sharing the fruits with humanity a powerful idea which I personally supported by may. Ravine had emailed with Facebook now Metta's Lecoon and corresponded with Nick Bostrom, a prominent and controversial Oxford University professor who spent decades writing about AI safety. With the help of Bernyolfson, he wrangled a brief in person with Stripe CEO Collison, and then with Andrew Ing, chief scientist at Baidu and one of the world's leading AI experts. Ravine then followed up with Collison by email, laying out the principles of open Ai. Collison said through a spokesperson that he had no record of the meeting and declined to comment about the email. Hung didn't reply to a request for comment, according to emails. Brynnolfson also told Ravine he'd pitched Ravine's ideas to a wider set of luminaries, including TED conference head Chris Anderson, LinkedIn co founder Hoffman, and Musk. I think we've made progress with Reed and others, he wrote to Ravine in late March. Hoffman declined to comment. Ravine believed he needed one hundred million dollars in funding to make open ai work, but the commitments were slow in coming. Ravine was not a guy who'd even come close to raising that kind of money before, and in the email exchanges he showed me it's difficult to discern genuine interest from a polite brush off. Sebastian Thrune, a former Google vice president and co founder of Udacity, begged off an advisory role, saying he was over committed. Sounds exciting, wrote Norvig, the head of research at Google, but by then the proposal had morphed into a physical school to teach AI concepts. Ravine was hustling, and like most people applying the valley for money, he was twisting and turning as he waded through the polite condescension. Still, he remained optimistic. That September, he added a splash page to open dot ai, declaring that an announcement will be made soon. Then on December tenth, he emailed Bostrum an update, musing on a future where no one group has an algorithmic advantage in general AI. What he was launching, Ravine hoped would shift the critical mass of development to an open effort. He had a feeling that the coming weeks could have a big impact on the course of evolution in a I. He was right, but not in the way he expected. For years, Ravine had a habit of keeping a strange kind of to do list, not of bucket list items or daily tasks, he says, but of like the things that most people don't do because they are insane, things like running fifty plus mile ultra marathons or swimming in San Francisco Bay every day for a year without a wet suit, so it was that on the afternoon of December eleventh, he jogged down to the San Francisco waterfront, stripped to his shorts and dived in. Emerging from the frigid water a few minutes later, he grabbed his phone from his shoe and noticed a one line e mail from Chris Cadieu, an AI researcher and entrepreneur with whom he'd been discussing open ai seen this. It was followed by a link to a blog post at OpenAI dot com announcing open Ai No Space, founded by Altman Brockman and Iliya Sutzkver, an esteemed AI researcher lured from Google. Open ai described itself as a nonprofit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. The post read to Ravine like it was ripped from his own proposals. His first thought was who are these guys? And then what the f The post asserted that open ai had already garnered a billion dollars in funding commitments from the likes of Hoffman and musk I started thinking, Okay, well, maybe it's a good thing, Ravine says, because the whole idea here was to create it, make it open for humanity. The vision was the idea. It wasn't about making money or my own ego. But there was another voice in his head, he says, one that said they stole the whole idea, the whole co and the name. Now at this point you might be thinking, sure, guy Ravine came up with a similar idea to open Ai. Okay, he even named it before they did and pitched it to some of the same people. But open Ai launched. Open Ai didn't game over. This is the unwritten code of the start up world, writ large. The winners are those who build, and by that same code, there's no bigger loser than the also ran inventor claiming their idea was stolen and clamoring for a piece of someone else's success. Think of the Winklevoss twins, enshrined in tech lore as chumps after Facebook settled with them for eight figures. Or Kevin Halpern, who in twenty fifteen sued Uber Technologies for a billion dollars, claiming that founder Travis Khalanik had stolen his idea for a cell phone powered car service after Halpern discussed in a shared office space. The suit wound on for seven years before a California court finally threw it out and sanctioned Halpern. There's Patrick rash a British inventor who spent decades entangled in an ongoing eighteen billion dollar claim against Apple, alleging in various courts that it stole his ideas for iTunes. Rash As said he's received death threats from Apple fans. My kid was being bullied, he told The Guardian recently, with kids saying your dad says he invented this. He didn't. He stole it. Apple invented it. How one views these stories reflects something of our feelings about genius ownership and ruthlessness in an industry that thrives on appropriation. But for those on the losing end, who believe they're watching their ideas make others wealthy and influential, it must feel like being frozen in time. They remain poised at the moment when the idea had promise, believing but never knowing they could have made it too. Ravine grew up in a small villa in Israel, where his mother ran the family's successful construction business. His early character fit the tech founder archetype, smart, precocious. Given a PC at age five, Guy was like my chat GPT, his brother, Saggie Reuven says, Ravine changed his last name after arriving in the US, sick of having to spell it for people. If I had a question, I would go to my mom and my mom would say, I don't know, go ask Guy. Ravine got into programming and created an online poker game that nearly got him kicked out of school. At eighteen, he managed to get himself excused from military service and moved to England. I didn't believe in what the Israeli government was doing, and also I thought it wouldn't be a good use of my time, is all he'll say about it. While taking classes at the University of Warwick, he made what he says was his first foray into AI, a chat bot capable of looking up answers on the Internet. It wasn't great, he says, and I thought it would take a long time to make it great, but it was good enough to get a small amount of funding and contacts to make his way to MIT, he audited classes and joined the startup scene, posting meetings for a tech incubator called MIT Playground on campus flyers and online message boards. A half dozen kids showed up, including future Cora CEO Adam Dangelo, then a California Institute of Technology student who was visiting the Boston area, and future collaborator McMurray, then a grad student. The group was fascinated with the ways fledgling social networks like Friendster and MySpace were struggling with speed as they scaled in size. Ravine had a company called I Need a Bit like if Task Grabit were also a social network that he says Dangelo also worked on. Guy had a sort of solution, an algorithm that dealt with the scaling issues, says McMurray. In Ravine's telling, he coded up the algorithm and sent it to Dangelo to implement, and then nothing. He disappears, never heard back from him. Ravine says, where does he emerge next? Co founder and CTO of Facebook responsible for scaling? What the f walking Ravine through his career is essentially a procession of these wtfs in which Ravine hatches ideas that later become major technology successes. In two thousand and two, it was a company that would let people amplify and share their Wi Fi connections that inspired what later became a billion dollar unicorn, he says. In two thousand and three, it was Stayway, a social network Ravine built that would allow people to rent out their spare rooms. He shows me a snapshot of the company's website, a map of the world covered in dots for shared lodging. It's going to be a multi billion dollar company, he remembers telling colleagues people were looking at me and scratching their heads. Airbnb made its debut five years later. In two thousand and seven, Ravine became convinced that electric vehicles were the future and designed his own single occupant electric car. He mixed in the ev community with the likes of early Tesla founders Martin Eberhart and Mark Tarpining, and wrangled contracts to build a prototype. The market collapse of two thousand and eight ended the dream of funding it. He did remain in touch with Eberhard and Tarpining, he says, and in twenty twelve I tried to convince them to invest a million dollars in bitcoin when bitcoin was ten dollars apiece. When I express some skepticism about this, Ravine pulls up the emails of him doing so. Eberhart and Tarpening didn't respond you requests for comment. He then shows me later ones after the pair failed to invest, in which he's ribbing them for missing out. Ravine himself did invest and eventually made a small fortune. Another example, he says, of seeing a future that others can't. It's not that Ravine always believes ideas were stolen from him. Sometimes he says he was just too early. Other times he couldn't line up the funding. Occasionally, though, there's some thread of connection that, at least to him, hints at inspiration, if not appropriation. Perhaps the singular example of Ravine's Zelig like technology career was his video sharing idea, the claim that had convinced me he might be a kook. In twenty eleven, he and McMurray were working on a dating app when they discovered that the friends they'd enlisted to test it were obsessed with one feature, the ability to share quick smartphone videos to your social network, as well as swiped through your friend's videos, which played instantly. They built a company around it called We. It aimed to capitalize on, as their funding deck put it, the massive opportunity of creating a purely video communication platform that is self contained. Ravine showed me old videos from WI, and I have to say, other than the fact that its users shot horizontally, it does look like Snapchat video before Snapchat video, or TikTok before TikTok. But among the reasons video had yet to take off was the cost of delivering it. Like you couldn't grow to one hundred thousand users and pay your bills and worry about making money later. McMurray says, you needed to be a high growth venture funded startup. When they scoured the valley for capital at firms such as Founder's Fund, though investors demanded tens of thousands of users, the same users they needed the investment to afford. Finally, according to Ravine, light Speed venture partners seemed ready to put up one million dollars. He says, one partner bulked, declaring that video sharing would never work, and the funding never materialized. As it happened, that partner was on the board of Snapchat now Snap Inc. Which added swipeable video sharing to its disappearing photo service within the year, TikTok emerged out of China a few years later. Neither company responded to inquiries, but facts which to Ravine feel suspicious even to Farius, to others simply amount to coincidence and a typical Valley funding rebuff. The light Speed partner reveal Vi says was ready to fund him, didn't respond to requests for comment. The one who Ravine says nixed the deal says it's just unequivocally not true that he said video wouldn't work. He doesn't remember Ravine, but says he invested in another social video platform at that same time. At any given moment, the world recognizes that a certain trend is coming, and there are multiple different inventions that emerge, Ravine says. He maintains that his implementation took off with those later companies because it was the most immersive and more accessible. McMurray moved on to other things, while Ravine obtained a patent for some of the technology and pivoted to a business to business video company called video dot Io, which was a bad idea, he says, but that's what I could get funding for. He believes Snapchat and TikTok are likely infringing on the patent, but says he hasn't had the stomach for legal action. Still, in the lawsuit with open Ai, his filings refer to him as having invented the foundation of modern mobile video, a field that was swarming with activity during that time. It's the kind of claim that can make Ravine seem well out over his skis video dot Io, which allows companies to make their own TikTok like apps, proved successful enough for Ravine to make comfortable living on top of his nest egg of prescient crypto investments. But there's comfortable and then there's what could have been. People who know me, my friends, they've always asked, how come you're not a billionaire, he says. They've been asking this for twenty years. They saw me coming up with all these things that people use today, and they're either stolen by other people or we couldn't get the funding for it. I don't know what to make of it. I could be depressed about it, or I could just live my life. We'll be right back with open Ai versus open Ai. Welcome back to open Ai versus open Ai. The day in December twenty fifteen, when open ai was announced, Ravine ran home from the beach. He took the first of several crucial steps that would set in motion the situation he finds himself in today. First, he went on the US Patent and Trademark Office website and filed for the trademark on open ai with a space. He did this without consulting a lawyer and lacking any real trademark expertise. And then at nine twelve p m. He sent an email to Altman and Sutzkever, two of open AI's three co founders. We've been working on an initiative called open dot ai to build a collective engineering platform to enable researchers from around the world to collectively engineer deep learning algorithms, he wrote. The initiative has the same goals as yours, which are to accelerate the arrival of general AI through an open effort. He mentioned some of the people he'd spoken to, Bostrom and Peter Norvig, with whom he'd discussed the AI school idea. It would make sense for us to meet and see if we could team up, he wrote, because we are working toward the same goals. This is bigger than us, and I think it could benefit from working together. Altman copied in Brockman, who suggested they meet at the Tesla offices in Palo Alto. When he arrived the next week, Ravine says, Brockman was immediately dismissive of any collaboration, saying Ravine should abandon the idea and sell them open dot ai. With open Ai already launched and funded to the tune of a billion dollars, what was the point in him continuing? Ravine says. Brockman emphasized that open Ai would be working for humanity's benefit, but to Ravine, he was acting like they just launched a hot for profit startup. The lack of empathy here is absolutely astonishing, he recalls, thinking, I don't know what I said, something like I'm not interested in selling, and then I walked out. Open Ai declined to make Altman or Brockman available for comment. Sutzkever didn't respond to request for comment. Feeling defeated, Ravine turned back to wiccaneering and his day job at video dot Io. I got into a depression over this for two years, he says, I thought, what is my purpose? Then still he continued to pursue the open ai trademark application. In twenty sixteen, he got a notice from the USPTO that it was being provisionally rejected, saying the screenshots of open dot ai he'd submitted failed to show it being used in commerce, a trademark requirement. Ravine submitted more screen shots, which he claimed showed an active discussion group on the site in twenty fifteen. Later in court, his lawyers would admit that the screen shots were from twenty sixteen, saying he'd lacked the records and had tried to recreate the earlier state of the site. The USPTO accepted them, and the trademark was granted in twenty seventeen, though not on the principal register but the supplemental one, because the name open ai was descriptive rather than distinct. Even with the trademark registration in hand, Ravine seemed to let the site languish. In another step that would prove fateful, he began having open dot ai redirect or forward to OpenAI dot com OpenAI's site. He left it that way, according to records submitted by OpenAI's lawyers, for four years. Ravine now says he never stopped working on public facing AI experiments, which he says were hosted at subdomains of open dot ai that never redirected, but looking back, it's easy to conclude he'd seeded the field to Altman and company. Then, in January twenty twenty two, OpenAI applied for its own trademark. Suddenly, Altman emailed Ravine for the first time since twenty fifteen, Following up on past conversations. He noted that open dot ai was forwarding to OpenAI dot com, then asked if Ravine would be open to us acquiring the open dot ai domain and related IP rights from you. This was the infamous Tesla email cited by open Ay's lawyers in their suit against Ravine, in which Ravine noted that Musk had paid eleven million dollars for the Tesla site and trademark, but open AI's complaint had conveniently excluded the rest of Ravine's response. The issue is that if you offered me a sum, I have no use for the money, he wrote to Altman. As an individual, I'm already well off. Ravine suggested that open ai donate money to a non profit academic AI research initiative he planned to organize. If it did, he'd hand over the domain and trademark. When Altman asked for specifics, Ravine replied that he was busy working to get my Russian team and their families out in time, in an operation that turned into a James Bond plot that keeps getting more insane. The Ukraine War had just commenced, and Ravine tells me he eventually managed to get his video dot io programmers out on the last flight out of Russia. Basically, sorry to hear, great that you are doing that. Altman replied that was their final correspondence. He could have easily responded and said, O K, how much do you want me to donate? Ravine says, now, we would have had a conversation and he would have donated probably a few million dollars at the end. That September, open ai created a sensation by fully releasing the latest version of its image generator doll E. Then, two months later, for reasons that are a central dispute in the lawsuit, Ravine re launched open dot ai with a new tagline, imagine if the best AI models were open and free. On the site, he embedded a version of an open source image generator stable diffusion. By December, it had one hundred and seventy thousand users, according to Ravine's court filings. The question was whose users were they. Ravine wrote a letter of protest to the U s p t O contesting open AI's trademark application on the basis that it would be confused with his. Then he started making plans to move to France and live in a room quote cabin away from it all for a few years. But Ravine had poked the bear, and as he packed up his house on August eleventh, twenty twenty three, he opened an email from a lawyer at the firm Quinn, Immanuel, Erkhart and Sullivan informing him that open ai was suing him in federal court over the domain and trademark. Bob Feldman, a partner at the firm who's litigating the case, declined to comment. I'm like what the f Ravine recalls Altman. He says could have had it for free, or at least the cost of a donation. Instead, he decided to donate millions of dollars to literally the most feared law firm in the world to sue me again and again. In our conversations, he returns to that phrase, the most feared law firm in the world. Finally, I ask him how he knows this. He turns his laptop toward me and pulls up the email. The signature reads Quinn, Immanuel, Erkhart and Sullivan Most feared law firm in the world. It's hard not to see a hint of irony in open AI's claims. This is a company facing a raft of lawsuits from publishers and writers accusing it of stealing copyrighted works to train its models. Open Ai has said its use of copyrighted material for training data is covered under fair use doctrine. Here it is in federal court accusing a single person of trying to profit off its identity and years of hard work. Open AI's initial complaint alleged that Ravine had infringed on open AI's trademark, a trademark the company had filed for but not received, in part because of Ravine's existing registration. Ravine, in the lawsuits telling was the latecomer, confusing the millions of users of open AI's products into mistakenly believing that open ai had some connection to open Ai. Defendant's motives are transparent. It alleged, namely to profit from their misappropriation of open AI's established name, leverage for themselves open AI's goodwill, or fraudulently divert public interest in and demand for open AI's products. At the heart of the case are questions raised by open Ai about who established the name open ai space or No in the market first, and whether Ravine's trademark registration is even valid, calling on expert witnesses, The company accused him of fabricating the USPTO evidence that he was using the mark in commerce. Ravine counters that he filed the application without a lawyer, misunderstood the USPTO's requests for samples of the site, and attempted to recreate faithfully how it looked when he filed. Judge Gonzales Rogers wasn't buying it. Her injunction barring Ravine from using open ai during the case seemed like a bad sign for Ravine's chances, at least in her court. Ravine, in his countersuit, accused open Ai of a different type of deception, not of the USPTO, but of him and the public. Despite the contention in its twenty fifteen blog post that Musk and the others had already pledged a billion dollars. According to its own financials, open ai raised just under fourteen million dollars in its first year. The billion dollar claim, Ravine said, tricked him into abandoning his own venture. The countersuit was stuffed with other claims, too, ranging from the trivial to the outlandish. At one point, Ravine's filing implied without evidence that D'Angelo, the former chief technology officer at Facebook now CEO of Cora and open ai board member whom Ravine had known in the early two thousands, had used old credentials to wipe evidence off ravine servers. A subsequent refiling removed the accusation. De Angelo didn't respond to requests for comment sent to Cora's press office, and open Ai declined to make him available for comment. As for how exactly open AI's founders had poached his idea, Ravine's claims were purely circumstantial. In the spring of twenty fifteen, two months after Ravine met briefly with Stripe's Collison, Brockman, announced he was leaving Stripe as CTO. In a blog post, Brockman noted that he sought advice at the time from both Collison and Altman. My goal was to figure out what deep learning was, he wrote, Whether Collison or some other luminary mentioned Ravine's ideas to Brockman or Altman seems difficult to establish, even if the suit makes it through discovery. Collison declined to comment, but Ravine's countersuit also rested on another argument that open ai has no right to the trademark because it isn't in fact open at all. Ravine's lawyer cited, among other events, the firing and re hiring of Altman last November, followed by the purge of open AI's board and the departure of its safety team. Open ai has undergone a remarkable transformation, Ravine's claim stated, from an open nonprofit organization with a figuciary duty to humanity into a closed for profit entity that appears to be under the de facto personal control of a single individual, Sam Alton. It's humanity's asset, Ravine tells me. It's not his asset. In Ravine's telling, this transformation from open to closed isn't the result of some considered change as open ai evolved, It's the natural product of the company's original sin Trying to build an idea they didn't really believe in that wasn't theirs to begin with. The countersuit went so far as to ask the court to force Altman to change its deceptive and misleading name to closed AI or a different, more appropriate name. This, as it happens, is the nearly identical contention of Musk in a federal lawsuit filed on August fifth, accusing open Ai, Altman, and Brockman of deceiving him into giving forty four million dollars to a non profit that isn't Once OpenAI Incs Technology approached Transformative AGI, Musk's suit contends Altman flipped the narrative and proceeded to cash in. In partnership with Microsoft, Altman established an opaque web of four per off at open ai affiliates, engaged in rampant self dealing, seized open ai inks board, and systematically drained the nonprofit of its valuable technology and personnel. Musk, a professed believer in the potential for superintelligent AI to doom humans, is no disinterested investor, having introduced his own chat GPT competitor grock He also has long claimed he came up with the name open Ai. In a strange twist, Musk's case was randomly assigned to Gonzales Rogers, the same federal judge as open Ai versus OpenAI. Musk's attorney didn't respond to an inquiry. In early October, open Ai raised six point six billion dollars at a one hundred and fifty seven billion dollar valuation and is reportedly discussing completing a for profit transformation and offering an equity stake to Altman. In court, open Ai moved to dismiss Musk's suit, calling it the latest move in Elon Musk's increasingly blusterous campaign to harass open ai for his own competitive advantage. But when Musk had filed his similar lawsuit in state court this year later withdrawn, OpenAI published a blog post purporting to show that musk knew of the company's intentions. Included was a private email from Sutzkiver to Musk in January twenty sixteen, suggesting that the open in open ai was temporary, meaning only that everyone should benefit from the fruits of ai A pledge to share the company's research openly, it continued, was merely the right strategy in the short and possibly medium term for recruitment purposes. It was never about being open, Ravine argues, noting that the email dates to just weeks after Brockmann had emphasized that open ai would be of public good. They come to me in person and make this representation, he says, But they already had these plans back then. This is how insane. This is insanity. And then there's the law, and in trademark law, says Genie Fromer, a law professor specializing in intellectual property at New York University, it's all about use in commerce. That's the whole ballgame. Unless Ravine can prove he established what's called secondary meaning for open ai, some amount of real users associating it with artificial intelligence products before open Ai did, it's unlikely to matter how open ai came about or how it's behaved, since even Ravine's supplemental registration amounts only to a kind of public placeholder. If you establish secondary meaning, Fromer says, you have the rights. In late September, Judge Gonzales Rogers again sided with open Ai, granting a motion to dismiss the bulk of Ravine's countersuit. She ruled that Ravine's lawyers hadn't established that open AI's for profit transformation was legally relevant. Ravine says he plans to refile an amended version. He's already appealing the judges a ration regional preliminary injunction against him. One thing is certain. It'll take months or years of legal rankling and likely millions more dollars for the case to see trial if Ravine and open ai settle. Instead, the questions the case raises about open Aye's mission and whether it was tainted from the beginning or abandoned along the way, could quietly disappear. The tech industry, despite its veneration of risk, isn't known for its courage in speaking truth to money, Ravine says. When he originally tried to hire lawyers to defend against open AI's suit, nearly thirty Bay Area firms turned him down, preserving their ability to work with open ai and Microsoft. He eventually found all his attorneys, including a now third firm hired weeks ago in Los Angeles, and except for Gruber, the people who heard Ravine's vision back in twenty fifteen have stayed silent when the lawsuit was filed, Ravine says, memories suddenly grew hazy and once enthusiasm stick boosters stopped returning his emails and mine. Brynnolfson, the man who helped shop Ravine's idea to the valley's elite, never responded to me. LeCun sent me to Metta's public relations team, who declined an interview on his behalf. Sorry, I didn't have any recollection of conversations like that, Norvig, replied when I emailed to ask about Ravine. When I noted that Ravine had shown me emails of their correspondence, he followed up, I do have a memory of meeting with someone on these topics. It must have been guy Ravine. I remember telling him to focus on what he wanted to achieve and to let the domain name or copyright or whatever on open Ai go. Bostrom, who'd once asked Ravine to read a draft of a paper and give notes, remembered him now only as a guy he ran into at a couple different conferences. He says, I don't recall much about what we were talking about, beyond a vague memory that he was telling me about some AI or AI compute related project or projects that he was pursuing or thinking about pursuing. After all, the events happened a decade ago, and even if some people do remember things Ravine's way, there's little advantage for them in playing even a bit part in a lawsuit involving a tech juggernaut backed by the most feared law firm on the planet. In one sense, all the parties in both lawsuits are united by a single belief that Obenai is building the most powerful technology in human history, an AGI that will change society forever. Both Musk and Ravine are nominally engaged in a fight to stop Altman from being the one who controls it, but that's not really something the court can do for them. Even if they win. Musk might get his money back, but too late to seriously wound his AI competition, and even if Ravine's version of events is vindicated, he's likely to remain unwelcome in Altman's AI future. The lawsuit is now a year in with a trial date continuously receding over the horizon. It's taken over his life, Ravine says, forcing him to lose focus on his own company, Video dot Io, which he'd been on the verge of selling. He stopped exercising and abandoned his insane activities lists. He used to joke around a lot. His brother, Saggi says, he has like this fun, sarcastic humor, and now it's like turned off, like hey, where's the fun guy. It's affecting him. Ravine spends much of his time alone in his house, helping lawyers prepare filings and answering discovery requests. It all amounts to a daily retreading of his life's work, searching for ways to sum it all up that'll prove he's not a fraud and a troll. It makes my heart kind of heavy. Saggie says, It's like we have a saying in Hebrew, how can you prove you don't have a sister? So is Ravine the little guy with pioneering innovations screwed out of history by more connected and ruthless operators, or the sore loos dreaming up ideas that he doesn't have the wherewithal to develop. By now, you've probably made up your own mind. But I wonder too if Ravine doesn't represent something else, a stand in for the rest of us facing a future that threatens to run us over, enriching an industry with no time for the thwarted dreamers and never made it. Build your own, they say, or get out of the way.