Altman, Musk and Hassabis back Amodei's call for independent oversight
OpenAI CEO Sam Altman, Elon Musk and former DeepMind CEO Demis Hassabis have all at least partly endorsed Anthropic CEO Dario Amodei's proposals for independent oversight inside AI labs, agreeing on the need for such oversight. Musk's support is not much of a surprise: he has been warning about AI risk for years, the report notes.
Altman separately told Fortune that OpenAI will not go public this year, citing safety concerns; he had already shared that decision internally in June. The report offers a second possible explanation for the delay: OpenAI's financials may simply not support an IPO right now, especially set against Anthropic's, and it notes that both reasons could be true at the same time.
Not everyone agrees with the reasoning behind Amodei's proposal. Google researcher Peyman Milanfar pushed back, arguing that recursive self-improvement (RSI) is a naive assumption because its feedback loops are inherently unstable: the harder a system optimizes itself, he says, the deeper it falls into blind spots, and reliable evidence about it comes only from the real world, not from benchmarks. That, in his view, makes Amodei's proposed "speed limit" unnecessary. "Systems that reliably improve themselves will be governed, damped, slow, and bounded by margins that look wasteful," he said. "Because stability is the speed limit."
Key facts
- OpenAI CEO Sam Altman, Elon Musk and former DeepMind CEO Demis Hassabis have at least partly endorsed Anthropic CEO Dario Amodei's proposals for independent oversight inside AI labs.
- The report calls Musk's backing unsurprising, since he has been warning about AI risk for years.
- Altman told Fortune that OpenAI will not go public this year, citing safety concerns, after sharing that decision internally in June.
- The report notes an alternative explanation: OpenAI's financials may simply not support an IPO right now, especially compared with Anthropic's, and says both reasons could be true at once.
- Google researcher Peyman Milanfar pushed back, calling recursive self-improvement a naive assumption because its feedback loops are inherently unstable, and arguing this makes Amodei's proposed "speed limit" unnecessary.
Why it matters
The story combines two separate signals about how seriously prominent AI figures are treating safety concerns right now. First, OpenAI CEO Sam Altman and former DeepMind CEO Demis Hassabis, alongside Elon Musk, have all at least partly endorsed Anthropic CEO Dario Amodei's call for independent oversight inside AI labs, agreeing on the need for it. That the position is echoed by leaders beyond Amodei himself gives it more weight than a single company's proposal would; Musk's support is the least surprising part, the report notes, since he has been warning about AI risk for years. Second, Altman told Fortune that OpenAI will not go public this year, citing safety concerns, a decision he had already shared internally in June. Together, the two moves suggest safety arguments may be shaping a concrete decision, not only public statements.
Who it affects
OpenAI is directly affected: an IPO this year is now off the table, which matters to anyone banking on a near-term OpenAI listing, on top of whatever reasoning actually drove the call. The endorsement also affects how Anthropic's push for independent oversight lands across the field: it is harder to dismiss as one company's self-interested pitch once OpenAI's Sam Altman and DeepMind's former CEO Demis Hassabis, plus Musk, are behind it too. And it affects the wider argument over how fast AI capabilities should be allowed to advance, since Google researcher Peyman Milanfar's pushback, that stability itself limits the pace, directly challenges the premise behind Amodei's proposal.
How to use it
There is no product or service here to use directly, so the practical takeaway is what to watch next. Anyone following AI governance should watch whether Amodei's proposal for independent oversight gains structure, authority, timeline or enforcement, none of which the report so far specifies. Anyone following OpenAI specifically should note that going public is off the table only for this year, not indefinitely, and that neither Altman nor the report settles which of the two explanations, safety or financials, is the real driver.
How solid is it
The central claim, that Altman, Musk and Hassabis endorsed Amodei's proposal, comes with a real hedge: the piece says they "at least partly endorsed" it, and gives no direct quotes or statements from any of the three showing how each of them actually backed it, only the assertion that they did. The IPO story stands on firmer ground, since it is attributed directly to Altman via Fortune and the report notes he had already shared the decision internally in June. Milanfar's rebuttal is at least quoted verbatim, more than the endorsement claim gets, but no calendar year is given for "June" or "this year" anywhere in the piece, so the timeline itself cannot be pinned down precisely.
Risks and caveats
The endorsements carry real limits: "at least partly" leaves open how much of Amodei's plan each of the three actually accepts, and since the piece specifies none of the practical details (who would enforce it, on what timeline, with what authority), the shared position could prove symbolic rather than a change to how any lab operates. Milanfar's rebuttal is a genuine counterweight to Amodei's premise, but the source gives no date, venue or forum for his remarks, so there is no way to check the context in which he made them. On the OpenAI side, framing the IPO delay purely as a safety call risks understating the alternative the report itself raises: OpenAI's financials may simply not support a listing right now, especially compared with Anthropic's, and the piece treats both explanations as possibly true at once rather than settling on one.
“Systems that reliably improve themselves will be governed, damped, slow, and bounded by margins that look wasteful. Because stability is the speed limit.”
— Peyman Milanfar, Google researcher