A group of current and former employees from OpenAI are advocating for their employers to eliminate non-disparagement agreements and permit whistleblowers to publicly voice their concerns about companies prioritizing growth and profit over safety. This initiative follows scrutiny of OpenAI CEO Sam Altman's investments outside the company by various media outlets. Seven former and four current OpenAI staff signed an open letter, arguing that due to AI companies' limited obligations to share information with governments and none with civil society, "current and former employees are among the few people who can hold them accountable to the public." However, extensive confidentiality agreements mean that whistleblowers risk losing their vested equity if they speak out. The letter also had two signers from Google DeepMind.
The signers called for AI companies to refrain from entering or enforcing agreements that prevent former employees from criticizing their previous workplaces. They advocated for creating a verifiably anonymous process for raising concerns to a company’s board and regulators, fostering a culture of open criticism (while protecting trade secrets), and not retaliating against whistleblowers who go public if their concerns are not addressed through official channels.
The signers expressed worry that AI companies are incentivized to avoid oversight and governance. Former OpenAI governance researcher Daniel Kokotajlo, one of the signers, told the New York Times that before joining OpenAI, he believed artificial general intelligence (AGI)—an AI capable of human-like cognition—wouldn't arrive until 2050. However, he now estimates a 50 percent chance the technology will arrive by 2027 and believes there's a 70 percent chance this advanced AI could destroy humanity. Conversely, figures like Meta chief AI scientist Yann LeCun have labeled those with significant concerns about AI development as "doomers" with a distorted view of reality.
Kokotajlo mentioned that he suggested to Altman last year that OpenAI should "pivot to safety" and allocate more resources to developing AI safeguards, instead of aggressively pushing new models and products. Although Altman supposedly agreed, little change ensued. On X, Kokotajlo explained why he signed the letter, stating that the lack of governmental AI oversight has left the public reliant on AI companies to self-govern, "even as profit motives and excitement about the technology push them to 'move fast and break things.'"
Despite Altman not having any ownership stake in OpenAI and earning only a $65,000 annual salary, he is a prolific investor with the Wall Street Journal reporting investments of at least $2.8 billion across more than 400 companies. These include a 7.6 percent ownership of Reddit and a $375 million investment in nuclear energy startup Helion, where he chairs the board. These companies have made deals with OpenAI that have significantly increased their valuations, benefiting Altman financially. His side projects and investments have come under more scrutiny since his brief ousting in November, with some directors arguing that Altman provided insufficient information about his startup holdings, complicating the understanding of how he might personally benefit from company deals. OpenAI stated that Altman didn't lead partnership discussions with Reddit, and he has recused himself from ongoing talks between OpenAI and Helion.