Former OpenAI CTO Mira Murati Testifies That Sam Altman Lied About Safety Review
Former OpenAI CTO Mira Murati's video testimony revealed that Sam Altman informed her that the legal team had approved skipping internal safety reviews for a new model, but after confirming with General Counsel Jason Kwon, she found their statements to be completely different.
Murati insisted on having the model pass through the safety committee review. She stated that Sam had undermined her authority as CTO, sowed discord among executives, and was not always honest; she believed the company was at "catastrophic collapse risk" at that time.
Former OpenAI board member Tasha McCauley testified that Sam presented three versions of ChatGPT to the board, but only one had truly gone through the safety process, calling it a "lie mode" that created a "culture of deception"; the board had significant concerns about Sam and multiple crisis incidents.
Source: Public Information
ABAB AI Insight
Mira Murati, a key builder of GPT-4 and briefly the CEO of OpenAI, focused her testimony on governance conflicts before and after Sam Altman's dismissal by the board in the fall of 2023, as the safety team was gradually disbanded (including the Super Alignment team), with core safety figures like Ilya Sutskever and Jan Leike leaving to start new companies.
On the capital front, OpenAI concentrated resources on commercial expansion through high-valuation financing (currently around $852 billion) and rapid product iteration (900 million users weekly), while systematically removing internal safety checks. Murati and McCauley's testimonies indicated that the board attempted to correct this but ultimately failed, and after Microsoft's intervention, Sam was quickly reinstated.
This mirrors historical paths like Purdue Pharma's rapid product promotion by suppressing internal warnings and the early governance crisis of Theranos. OpenAI is currently in the later stages of transitioning from founder control to high commercialization, with safety culture significantly weakened and remaining safety advocates largely exited.
Structural judgment: This essentially involves a transfer of pricing power. Sam Altman concentrated control and decision-making speed on commercial expansion by weakening internal safety reviews and governance checks, driven by the capital market's extremely high premium on growth and user scale, far exceeding the pricing of long-term risks, pushing resources from safety checks towards rapid iteration and lobbying capabilities of a few leaders.
ABAB News · Cognitive Law
When safety reviews are bypassed, the higher the valuation, the more risks are underestimated.
The stronger the internal warnings, the more external financing packages the issues as "innovation."
Whoever can turn governance crises into growth fuel temporarily holds pricing power, but history ultimately settles the accounts.