Sam Altman, Elon Musk, and Demis Hassabis have endorsed Dario Amodei’s proposal to introduce independent oversight into artificial intelligence laboratories. This alignment between the leaders of OpenAI, xAI, and DeepMind signals a shift toward slowing development speeds to mitigate safety risks. Altman confirmed to Fortune that OpenAI will not pursue an initial public offering this year, citing safety concerns as the primary driver. While financial constraints may also play a role, the decision highlights the tension between rapid commercialisation and responsible deployment.
The consensus on oversight contrasts with dissenting views from researchers like Peyman Milanfar. Milanfar argues that the assumption of recursive self-improvement is flawed because feedback loops become inherently unstable during optimisation. He contends that systems governing themselves will naturally be bounded by stability margins, making artificial speed limits unnecessary. The debate centres on whether external checks are required or if internal system dynamics will inherently prevent runaway acceleration.
* Altman previously shared the decision against an IPO internally in June
* OpenAI’s current financial position differs significantly from Anthropic‘s
* Milanfar believes stability acts as a natural speed limit for self-improving systems
