Sam Altman, chief executive of OpenAI, sat down for an hour-long interview in San Francisco to address growing public alarm over artificial intelligence. The conversation came after an employee at rival lab Anthropic warned on X that AI could wipe out humanity, echoing long-standing concerns voiced by figures such as Elon Musk and Anthropic founder Dario Amodei.
Assessing the risk
Altman said the probability of an existential AI disaster, often referred to as "p-doom", is not a theoretical exercise. He noted that Musk has estimated a 10-20% chance, while Amodei has placed his estimate between 10% and 25%. "No gamble with humanity is OK," Altman told me, underscoring his belief that the stakes are too high for complacency.
Regulatory and industry response
The CEO argued that national and international regulators must act swiftly, pointing to the European Union's AI Act as a model for coordinated oversight. He hinted at a forthcoming pact among leading AI firms to deliberately slow the rollout of more powerful models until safety measures catch up.
IPO postponed until 2027
When asked about OpenAI's planned public offering, Altman described it as "ill-timed" given the safety concerns. He confirmed that the IPO will not take place before 2027, allowing the company to focus on alignment research.
What comes next?
Altman said he would be prepared to pause or even stop AI development if it became clear that safety could not be guaranteed, even if that meant confronting investors. The interview suggests a shift towards caution in the fast-moving AI sector, with potential ripple effects for European policy makers and investors alike.