Democrats press Sam Altman on OpenAI security

“Given the discrepancies between your public comments and the reports of OpenAI’s actions, we request information about OpenAI’s whistleblower protections and conflicts of interest to understand whether federal intervention may be necessary,” Warren and Trahan wrote in the report. letter made available exclusively Edge.

The lawmakers cited several cases that questioned OpenAI’s security procedures. For example, they said, in 2022, an unreleased version of GPT-4 was tested in new version of Microsoft Bing search engine in India before receiving approval from OpenAI’s security board. They also recalled Altman’s brief ousting from the company in 2023, over the board’s concerns, in part, “about commercializing advances before understanding the consequences.”

Warren and Trahan’s letter to Altman comes as the company has been plagued by a long list of security concerns that often contradict the company’s public statements. For example, an anonymous source he said The Washington Post that OpenAI rushed security testing, the Superalignment team (which was partly responsible for security) was disbanded, and the chief security officer resigned, saying that “security culture and processes took a back seat to shiny products.” Lindsey Held, a spokeswoman for OpenAI, denied those claims in The Washington Postand that the company “did not skimp on our security process, although we recognize that the launch was stressful for our teams.”

Other lawmakers also wanted answers about the company’s security practices, including a group of senators under Chairman Brian Schatz (D-HI) in July. Warren and Trahan asked for further clarification OpenAI’s responses to this groupincluding the creation of a up-to-date “Integrity Hotline” through which employees can report their concerns.

Meanwhile, OpenAI seems to be on the offensive. In July, the company announced a partnership with Los Alamos National Laboratory to explore how advanced AI models can safely aid bioscience research. Just last week, Altman announced via X that OpenAI was partnering with the U.S. AI Safety Institute and emphasized that 20 percent of the company’s computing resources would be dedicated to safety (a promise originally made to the now-defunct Superalignment team). In the same post, Altman also said that OpenAI had removed employee nondisparagement clauses and provisions that allowed for the cancellation of vested shares, a key point in Warren’s letter.

Warren and Trahan asked Altman for information about how the up-to-date AI safety hotline for employees is being used and how the company is tracking reports. They also asked for a “detailed accounting” of all instances in which OpenAI products “bypassed security protocols” and under what circumstances a product would be allowed to bypass a security review. The lawmakers are also seeking information about OpenAI’s conflicts of interest policy. They asked Altman whether he was required to divest from any outside holdings and “what specific safeguards are in place to protect OpenAI from financial conflicts of interest?” They asked Altman to respond by Aug. 22.

Warren also notes how vocal Altman has been about his concerns about AI. Last year, Altman spoke before the Senate warned that the capabilities of artificial intelligence could be “significantly destabilizing to public safety and national security,” and underscored the impossibility of predicting every potential misuse or failure of the technology. Those warnings seemed to resonate with lawmakers — in OpenAI’s home state of California, Sen. Scott Wiener is pushing a bill to regulate gigantic language models, including restrictions that would hold companies legally liable if their AI is used in a harmful way.