Table of Contents
ToggleNonprofit organization reveals the ugly side of ChatGPT's parent company OpenAI
Rob Wiblin is the program director and host of The 80,000 Hours Podcast , a podcast that explores the world’s most pressing issues through in-depth conversations from different perspectives. The show has also invited Vitalik Buterin to discuss topics such as AI regulation.
Rob Wiblin shared the " Open AI Document " on Twitter. The document was compiled by The Midas Project and The Tech Oversight Project. It is the most comprehensive public record of OpenAI's governance practices, leadership integrity, and organizational culture to date. The document's sources include speeches by former employees who have left OpenAI, media reports, etc.
The Midas Project is a nonprofit watchdog group that aims to ensure that AI companies are socially responsible, while the Tech Oversight Project is holding big tech companies accountable for anti-competitive practices and internal corruption.
Sam Altman's eight deadly sins: lying openly to the SEC and Congress?
Rob Wiblin handed the document to Claude for sorting out, and found several major omissions:
Did Sam Altman lie to the SEC about being the chairman of Y Combinator?
The Wall Street Journal noted that before stepping down as president of Y Combinator, Altman proposed to promote himself from president to chairman in order to smoothly leave. He preemptively published a blog post on the company's website announcing the change. But the company's partners never agreed, and the announcement was later removed from the article. But Altman still falsely claimed that he was chairman in documents submitted to the U.S. Securities and Exchange Commission (SEC) for many years.
Sam Altman indirectly holds shares in OpenAI through Sequoia Capital and YC
Sam Altman claimed to Congress that he did not have any equity in OpenAI, but he indirectly held shares in OpenAI through Sequoia Capital and Y Combinator. At the same time, Sam Altman did not disclose to the board of directors that he owned the OpenAI startup fund for many years. He required board members to inform every time they talked to employees, which limited the board's oversight.
OpenAI has an equity recovery clause, if the former employee criticizes the company, the vested equity will be recovered
Sam Altman denied that he knew about the existence of equity clawback clauses, which threaten to deprive departing employees of millions of vested shares if they criticize OpenAI. But Vox found that Altman personally signed documents authorizing these clauses in April 2023. These restrictive nondisclosure agreements prohibit employees from even acknowledging the existence of the agreement.
Sam Altman did not comply with the principle of conflict of interest, OpenAI cooperated with its investment units
Sam Altman did not comply with the recusation rules. He owns 7.5% of Reddit, and when Reddit announced its partnership with OpenAI, Altman's net worth jumped by $50 million. Sam Altman also invested in Rain AI, and OpenAI subsequently signed a letter of intent to purchase $51 million worth of chips from it.
Claude founder accuses Sam Altman of emotional manipulation and psychological abuse in running the company
Dario Amodei and Daniela Amodei, founders of Claude's parent company Anthropic and former OpenAI employees, described Sam Altman's management strategy as "gaslighting" and "psychological abuse." At least five other OpenAI executives have provided similar negative feedback about Altman to the board of directors.
OpenAI was stolen but kept quiet, fired whistleblower
OpenAI suffered a major security breach in 2023, when a hacker stole AI technical details but did not make it public for more than a year. OpenAI fired Leopold Aschenbrenner because he expressed security concerns to the board of directors.
OpenAI executives also don’t trust Sam Altman to lead the product toward AGI
Ilya Sutskever, co-founder of OpenAI, told the board: "I don't think Sam is capable of leading AGI (artificial general intelligence)." Chief Technology Officer Mira Murati also said: "I am uncomfortable with Sam leading us towards AGI."
OpenAI secretly modified its profit cap but still prides itself on being a limited-profit company?
OpenAI's profit cap was quietly adjusted to 20% per year. At this rate, its profits will exceed $100 trillion in 40 years. This adjustment was not made public, and OpenAI still prides itself on the company's profit cap structure, but does not admit to quietly modifying the cap.
Musk blasts Ultraman as a fraud
Elon Musk also reposted the article and wrote : "Sam Altman is a fraud." In fact, Elon Musk was one of the original board members of OpenAI, but left in 2018 because he was worried that the development of GPT-4 and the next generation of AGI might endanger public safety. He believes that Microsoft's control over Sam Altman and the OpenAI board will prevent GPT-4 from transforming into AGI, thereby maintaining the confidentiality and profitability of related technologies.
(Musk resumes lawsuit against OpenAI and Sam Altman: using non-profit slogan to lure investors )
He criticized OpenAI for being only concerned with making money, so he filed a lawsuit against OpenAI and Sam Altman, demanding that the product return to open source.
( Musk sues OpenAI and Sam Altman for only making money, demanding a return to open source )
Rob Wiblin also said that the above is only part of the tens of thousands of words of the file. He skipped the question about the reorganization of OpenAI, but that is the real bigger issue.
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
In the fiercely competitive artificial intelligence battlefield of OpenAI, Google, and Anthropic, Meta often appears low-key and marginal. Although it has its own open source models, it has always lacked breakthrough applications or innovations. Now, Meta's repeated large-scale mergers and acquisitions and poaching actions are showing that Meta seems to have shifted from focusing on research and development to using capital and ecological layout to make up for the lagging market presence, hoping to regain the dominance of the AI narrative.
Table of Contents
ToggleFrom R&D LLM to large-scale mergers and acquisitions: Meta's strategic shift
In the past two years, Meta has devoted its efforts to developing the LLaMA series of language models. Although Llama 4 has been launched, the market response has been lukewarm. The Wall Street Journal (WSJ) said that developers generally believe that rival models such as Gemini 2.5 Pro, Claude 4 and GPT-4o are stronger, not only in terms of reasoning capabilities, but also more like humans.
In addition, the successive job-hopping of core members has also caused the outside world to worry about the development prospects of Meta AI.
Compared with market competitors, what is the difference between LLaMA?
The author believes that the main problem is that "good technology does not mean a good product, nor does it mean it meets market demand." An excellent model with excellent performance and open source freedom actually lacks the ability to build user habits and dependence and integrate with a practical product.
OpenAI relies on ChatGPT to establish a user portal; Google uses Gemini to build the Android ecosystem; Anthropic focuses on enterprise security and API services. Although Meta has integrated LLaMA into its own social software empire Instagram or WhatsApp and other products, it is obvious that the user usage rate and market popularity are both low.
Meta CEO Mark Zuckerberg obviously realized that product development alone could not support the full competition of the AI ecosystem. Therefore, a new layout of mergers and acquisitions and investments gradually unfolded.
If you can’t win, join: Meta acquires Scale AI and NFDG
Since 2024, Meta has launched a series of proactive attacks. First, it invested US$14.3 billion in data labeling giant Scale AI, acquired nearly half of its shares and recruited its founder Alexandr Wang to join Meta's newly established "Superintelligence" department.
It then attempted to acquire Safe Superintelligence (SSI), an AI company founded by OpenAI co-founder Ilya Sutskever. Although the attempt was rejected, it successfully poached SSI CEO Daniel Gross and former GitHub CEO Nat Friedman, and acquired shares in NFDG , a venture capital fund they jointly founded.
This is not just a talent acquisition, but also a strategic ecological penetration. NFDG's investment targets not only have outstanding performance now but also have promising future, including Figma, Stripe, Perplexity and Eleven Labs.
This transaction allows Meta to indirectly obtain these newly created technology resources, investment opportunities and acquisition paths, and enter key application scenarios at a lower cost in the future.
Talent magnet + data collection: Zuckerberg's multiple AI layouts
Dragonfly investor Omar wrote an article pointing out the key to this deal:
Zuck uses one investment to combine industry intelligence, strategic layout, brand enhancement and talent absorption.
Gross and Friedman are not only experienced entrepreneurs, but also "talent magnets" in the entrepreneurial circle. Their joining means that Meta will be more likely to attract top AI engineers, product designers and startup teams.
Simply put, investing in Scale AI has enabled Meta to strengthen its control over data, while binding with NFDG has filled the talent gap at the application layer.
From a lone army to an alliance, Meta uses money to infiltrate the AI ecosystem
Meta and Zuckerberg now seem to have shifted from incubating the AI ecosystem internally to forming an AI landscape that spans products, companies, and talents through alliances based on capital and connections.
Whether it is open source models, talent acquisition, investment or acquisition, Zuckerberg is using a mixed strategy to fill the company's "presence gap" in the AI market. While other technology giants continue to strengthen their application control and market share, Meta chooses to spend money to form alliances, hoping to gain the key to victory in this AI competition.
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
