Whistleblowers Reportedly Accuse OpenAI of Enforcing ‘Illegal’ NDAs

OpenAI has reportedly been accused of using “illegally restrictive” non-disclosure agreements (NDAs) by a group of whistleblowers, who have written a letter to the US Securities and Exchange Commission (SEC), urging the agency to take “swift and aggressive” steps to stop the practice within the company. Last month, a group of anonymous former and current employees of the company wrote an open letter highlighting the lack of oversight in building safe artificial intelligence (AI) models and urging employers to create better whistleblower protection policies.

Whistleblowers Says OpenAI NDAs Are Restrictive

The Washington Post obtained a letter from a group of anonymous individuals describing themselves as whistleblowers, addressed to SEC chair Gary Gensler. The letter states that OpenAI has required its employees to sign illegally restrictive NDAs. The letter mentions that these NDAs prohibit and discourage both employees and investors from communicating with the SEC about security violations.

The letter alleged that the NDA requires both employees and investors of the company to obtain consent before reaching out to federal agencies. Whistleblowers have reportedly also sent a separate, formal complaint about OpenAI along with documents as evidence of the company’s wrongdoing.

In the letter, the group urges the SEC to begin an investigation into the matter. A spokesperson for Senator Chuck Grassley told TechCrunch that the letter was received by the SEC and copies of it were sent to Congress.

The whistleblowers also said that there is an urgent need for intervention, claiming there are severe risks attached to building AI systems and the irregularities at the company required that the employees could communicate with federal institutions. It could not be ascertained whether the whistleblowers are currently working at OpenAI, or whether they were former employees.

Last month a group of former and current OpenAI employees wrote an open letter warning about the perils of AI and the need for better scrutiny when building such technologies. The letter also highlighted that the oversight into the matter was negligible and the employees needed better whistleblower protection policies to be able to come out and speak freely. Recently, a separate report also claimed that OpenAI employees found the safety and security protocol for GPT-4o to be insufficient.