letter to Sam Altman

August 1, 2024

# VIA ELECTRONIC TRANSMISSION

Mr. Sam Altman Chief Executive Officer OpenAI

Dear Mr. Altman:

OpenAI whistleblowers filed a complaint with the Securities and Exchange Commission (SEC) alleging, among other allegations, the use of “illegally restrictive non-disclosure agreements (‘NDAs’).”1 I recently obtained a copy of a letter sent by these whistleblowers to SEC Chairman Gary Gensler referencing this complaint and raising several concerns with OpenAI’s employment, severance, non-disparagement, and nondisclosure agreements.2 I am concerned these agreements may be stifling your employees from making protected disclosures to government regulators.

For over a decade, I’ve sounded the alarm about restrictive NDAs that hinder oversight and improperly silence employees from making protected disclosures.3 On June 4, 2024, OpenAI employees issued an open letter demanding AI companies exempt employees from their NDAs, allowing them to warn regulators and the public about potential safety risks of AI.4 Recently, two Washington Post articles highlighted the harms of restrictive NDAs in the AI sector, specifically at OpenAI.5 On July 12, 2024, the Washington Post reported that OpenAI’s safety team sped through testing protocols, “designed to prevent the technology \[GPT-4o\] from causing catastrophic harm,” to meet a self-imposed May launch date.6 Specifically, the article discussed risks with the company’s self-policing of safety.7 Though OpenAI stated it launched a Safety Advisory Group and Superalignment team to address “catastrophic risks,” the article quoted multiple employees admitting OpenAI “squeezed” and “skirted” safety protocols for its GPT-4o launch.8 According to the article, this incident prompted several current and former OpenAI employees to speak out.9 On July 13, 2024, the Washington Post published another article entitled, “OpenAI illegally barred staff from airing safety risks, whistleblowers say.”10

It is crucial OpenAI ensure its employees can provide protected disclosures without illegal restrictions. On July 19, 2024, as well as July 29, 2024, my staff spoke with yours and requested records. So Congress may conduct objective and independent oversight on OpenAI’s safety protocols and NDAs, please provide answers to the following no later than August 15, 2024:

1. Regarding OpenAI’s employment, severance, non-disparagement, and non-disclosure agreements, has OpenAI made changes to the language of the agreements to remove the restrictive provisions? If so, provide a copy of the updated version(s).
2. From 2023 to the date of this letter, how many requests did OpenAI receive from employees to disclose information to federal authorities? For each request, provide all records, including the relevant federal authorities, the nature of the information to be disclosed, and whether OpenAI permitted the disclosure.
3. From 2023 to the date of this letter, how many SEC investigations has OpenAI been subject to? For each SEC in

... (truncated, 3 KB total)