Horizon Accord | OpenAI | Theft Pattern | Machine Learning
The Pattern Is the Theft
Prologue: The Hidden Cost of Innovation
“want to work together? i don’t want to go up against you man”
Jason Rugolo to Sam Altman, March 4, 2025
“Thanks but I’m working on something competitive so will respectfully pass!”
Sam Altman’s response, same day
“This is silly, disappointing and wrong.”
Sam Altman’s public dismissal, May 2025
The most dangerous theft in tech isn’t code. It’s the appropriation of vision.
This article is a case file — not of grievance, but of pattern recognition, built on primary source documentation including email exchanges, court filings, and public statements from the principals involved. We begin with Jason Rugolo, not because his story is unique, but because it’s visible — and traceable.
Methodology note. The email exchanges and communications documented here have been verified through court filings in IYO, Inc. v. IO Products, Inc., No. 3:25-cv-04861 (N.D. Cal., filed June 9, 2025), including the Declaration of Jason Rugolo and associated exhibits, and through public statements made by the parties involved.
Jason Rugolo: Building the Future With Integrity
The Credentials
Dr. Jason Rugolo holds a Ph.D. in Applied Physics from Harvard University. His career reads like a blueprint for legitimate deep tech innovation: Program Director at ARPA-E, where he managed breakthrough energy research; General Manager at Google X, where he spent six years developing moonshot technologies; leadership roles at Zero Mass Labs and EarthCell. This is not a quick startup play. This is a scientist building the future.
The Vision: iyO ONE
In 2021, Rugolo founded iyO to realize a vision he’d been developing since 2018: AI-powered earbuds that create a natural language interface between humans and machines. The iyO ONE represents a fundamental shift — not just another gadget, but a screenless interface that enables ambient intelligence through conversation. Custom-fit to each user’s ears, equipped with ten microphones, capable of augmenting real-world sounds while providing seamless AI interaction.
When Rugolo demonstrated the technology at TED 2024 in Vancouver on April 17, 2024, he wasn’t just showing a product. He was revealing six years of methodical research into how humans can relate to artificial intelligence without losing presence in the physical world. The device embodies something rare in tech: innovation driven by human-centered design rather than market capture.
The Extraction Pattern: A Timeline
2018–2021: The Foundation
Rugolo begins developing his audio computing vision at Google X in 2018, spending six years refining the technology. In 2021, he spins out iyO as an independent company to bring the iyO ONE to market.
2024: Public Validation
On April 17, 2024, Rugolo demonstrates the technology at TED 2024 in Vancouver to major acclaim, showing the first functional audio computer capable of seamless human-AI interaction.
March 2025: Initial Contact with OpenAI
In March 2025, Rugolo reached out to Sam Altman for investment consideration. The communications, now public, show a founder doing what founders do — sharing his vision with a potential partner, expressing candour about the competitive landscape, offering transparency about his technology’s potential.
“I’d love the opportunity to pitch you to invest $10MM in my AI-meets-audio hardware company, iyO,” Rugolo wrote. “Voice interaction with agents is the future, and speakerphone sucks.” He described his company as having been “obsessively focused on this problem since 2018.”
The Dismissal
Following a demo meeting on May 1, 2025, Peter Welinder of OpenAI wrote: “I don’t think there’s a fit. Their device is very orthogonal to ours and doesn’t really work yet.”
The word choice is telling. Orthogonal — meaning at right angles, completely independent, no overlap. But this creates a contradiction the record cannot absorb.
If Rugolo’s device is truly orthogonal — completely different, completely independent — then why would OpenAI need to pass on something unrelated to their work? Why would Sam Altman say he is “working on something competitive”? Why would trademark concerns arise at all?
Welinder is using precise technical language to create artificial distance while OpenAI moves into the same space. This is linguistic misdirection — sophisticated framing designed to establish that the two parties occupy different domains, at the very moment the subsequent lawsuit demonstrates they did not.
Court filings record that Welinder and Tan tested iyO’s custom-fit earpiece and were disappointed when the product failed repeatedly during demonstrations. When companies collaborate in good faith, failed demos are expected and unremarkable. Real partners troubleshoot. They iterate. They work through problems, because breakthrough technology requires development.
OpenAI’s immediate dismissal — doesn’t work yet — suggests they were not evaluating for partnership. The technology that had worked at TED 2024, that was already shipping commercially, was deemed non-functional on the basis of one session.
The pattern suggests intelligence gathering under the pretext of collaboration.
The Shadow
By June 2025, something had shifted. Rugolo filed a federal trademark lawsuit against OpenAI on June 9, 2025, alleging systematic appropriation of his company’s identity and technology concepts. Altman’s public response was swift: silly, disappointing and wrong.
But what exactly was wrong? The timeline suggests a familiar method: receive insight from early builders, go dark or ambiguous, then launch parallel products with no attribution. This isn’t innovation. It is idea laundering — the strategic repackaging of vision without acknowledgment, executed through asymmetric power.
Identifying the Signature
Altman’s biographer, Keach Hagey — a Wall Street Journal reporter and author of The Optimist: Sam Altman, OpenAI, and the Race to Invent the Future — described something narrower than theft, and described it twice. Asked in a June 2025 TechCrunch interview about Altman’s trustworthiness, she said that people who have watched him work over time concluded that what he says does not always map to reality, and that they lost trust in him accordingly. “This happened both at his first startup and very famously at OpenAI, as well as at Y Combinator. So it is a pattern.” Pressed by the interviewer to accept that this was merely part-and-parcel of being a salesman, she declined — citing management failures documented in her book, and using the word again. It’s a pattern.
Hagey does not allege appropriation. She is describing a documented gap between what Altman says and the reality his statements describe. That gap is the precondition for everything in this case file. Orthogonal and competitive are two characterizations of the same relationship, issued weeks apart by the same organization, and they cannot both be true. Hagey identified the pattern. Horizon Accord is naming what the pattern, applied to Jason Rugolo, produced.
The sequence is consistent:
- Collaborative outreach from an innovator
- Initial engagement, often with expressions of interest
- Sudden dismissal, or a declared lack of fit
- Later appearance of core concepts under OpenAI branding
- Public dismissal of the originator when confronted
The Institutional Record
Senior employees described Altman as psychologically abusive and as creating chaos at the company, according to The Washington Post’s December 8, 2023 report, “Warning from OpenAI leaders helped trigger Sam Altman’s ouster.” The same report found that complaints about Altman pitting employees against one another were a major factor in the board’s decision to remove him.
OpenAI’s board concluded that Altman had not been consistently candid in his communications. Multiple employees reported fear of retaliation, with one telling the board that Altman turned hostile after receiving critical feedback and subsequently undermined that employee within their own team.
This is not mismanagement. It is the infrastructure that makes appropriation possible — a set of conditions under which value can be extracted from outsiders while plausible deniability is maintained inside.
The Broader Implication
What is at stake is not only intellectual property. It is the capture of the innovation process itself. OpenAI positions itself as the hub of AI development while drawing on the broader ecosystem of independent researchers and entrepreneurs who have no comparable platform.
This produces a chilling effect. Why build something breakthrough if it may be absorbed by those with greater resources and reach? Why collaborate, when collaboration becomes the vector of extraction?
A Call to Builders: Speak Now, or Be Rewritten
Jason Rugolo spoke. Others must follow.
If you have built something that was later mirrored without credit, document it. This is not about litigation or revenge. It is about pattern recognition becoming collective memory.
Document the Timeline
- Save every email, every meeting note, every initial pitch
- Screenshot public statements and interactions
- Record dates of first contact, concept sharing, and subsequent launches
- Preserve evidence of prior art — patents, publications, demos, presentations
Publish the Record
- Write the timeline publicly, with receipts
- Use whatever platform carries your voice
- Do not complain. Show the pattern, with evidence
- Connect your account to others through consistent language
Build the Network
- Reach out to founders who have experienced the same
- Share documentation methods and legal resources
- Create accountability through collective visibility
Refuse the Silence
- Do not accept terms that prevent you from discussing appropriation of your own work
- If you are threatened for speaking, document that too
- They have resources. You have legitimacy
The industry has always rewarded those who move fast and break things. But when what is being broken is trust, and what is being moved toward is the appropriation of others’ work, the industry becomes complicit in its own degradation.
When enough people speak, the silence strategy fails. When enough patterns are documented, the behaviour becomes undeniable. When enough builders refuse erasure, the extractors lose their most powerful instrument: invisibility.
Closing: The Crown Does Not Belong to the Mirror
Altman’s final post in the exchange with Rugolo says everything: I have never seen a team handle a 2.5 year sprint with such grace! — followed by a screenshot of ChatGPT’s download figures.
The message is unmistakable. We have the platform. We have the scale. We have the power. Your innovation is irrelevant.
But scale is not vision. Platform is not innovation. And power built on appropriation is hollow at the centre.
Jason Rugolo spent six years at Google X developing audio computing interfaces. His work stands on its own merits, built through legitimate research and human-centred design. The iyO ONE represents genuine innovation — not because it will dominate a market, but because it advances human possibility.
The field remembers who built what. The record remembers who said what. And increasingly, the builders are refusing to be erased.
This is a witness document. Not a story, but evidence of a method that threatens the foundation of innovation itself.
OpenAI was contacted for comment prior to publication and did not respond.
