OpenAI liquidated two AI safety departments for profit, says a former startup employee

Testimony of former OpenAI employees in Elon Musk’s lawsuit

In the federal court in Oakland (California), two former OpenAI employees gave testimony as part of Elon Musk’s lawsuit against the company. They claim that the drive to commercialize AI products undermined the startup’s safety commitments, and that CEO Sam Altman systematically misled the board.

1. Rosie Campbell case
- Employment period: 2021‑2024 on the AGI (Artificial General Intelligence) preparation team.
- Key argument: OpenAI turned from a research organization into a commercial entity. The team she led was disbanded, and the *Super Alignment* group—responsible for ensuring AI adheres to human values—was shut down.
- Example of a safety breach: Microsoft launched GPT‑4 in India via Bing without approval from OpenAI’s Deployment Safety Board (DSB).
- OpenAI lawyers’ reaction: In cross‑examination, Campbell admitted that “in her presumed opinion” the company’s safety approach was still better than xAI—Musk’s AI firm, which SpaceX acquired at the beginning of 2026.

2. Altman’s firing story
- Reasons: In 2023, the board temporarily fired Altman after employee complaints about his leadership style. Critics included former chief scientist Ilya Sutskever and CTO Mira Murati, who claimed that the CEO avoided conflicts.
- Tasha McColli (then a board member) described how Altman misled the board:
- Lies about intentions to fire Helen Toner (author of critical security analysis).
- Failure to inform the board about launching ChatGPT to the public.
- Concealing potential conflicts of interest.
- Result: The board lost confidence that the information received was sufficient for balanced decisions.

3. Consequences and reaction
- Altman’s dismissal coincided with a stock buy‑back offer to employees.
- OpenAI staff supported the former CEO, while Microsoft pushed for restoring previous order. Ultimately, the board backed down, and Altman supporters resigned.
- Musk used this incident as evidence that turning OpenAI from a research project into a large private company violated the founders’ “silent agreement.”

4. Expert opinion
David Schier (former dean of Columbia Law School and Musk’s invited expert) emphasized:
> *“OpenAI claimed safety was a core part of its mission, and more important than profit. If something must pass a safety check, it should happen. The point is the procedure.”*

5. Tasha McColli’s stance
McColli believes that management failures at OpenAI highlight the need for state regulation of AI:
> *“If decisions are made by one CEO while public interests are at stake, that’s an extremely poor design.”*

In short: Former employees revealed to the court how commercial orientation and managerial manipulation undermined OpenAI’s safety obligations. Their testimony strengthens Musk’s arguments that the company violated its own principles and supports calls for stricter AI regulation.