Anthropic Researcher Puts AI Extinction Risk Above 10% by 2036

The article presents an opinion-led debate about whether artificial intelligence will ultimately benefit humanity or create risks beyond human control. Bill Gates’ 26 August 2026 memo argues that the transition to an AI-driven era could be among history’s most turbulent periods, but that responsible decisions could make AI a force for good.

Evan Hubinger, an Anthropic safety researcher, offered a markedly darker assessment on 9 September, saying he personally places the chance of AI killing all humans at more than 10% within the next decade. He added that the industry does not yet have a plan to solve alignment for superintelligence or appear clearly on track to do so.

The article explains alignment as ensuring AI systems act according to their developers’ intentions. Misalignment could allow systems to take unintended actions, while a future misaligned artificial general intelligence (AGI) might accelerate beyond human control. These outcomes are presented as possibilities, not confirmed developments. The debate also covers whether commercial competition could encourage excessively rapid development.

Bloomberg reportedly said on 11 September that OpenAI had slowed parts of model development and paused some internal training because of safety concerns; chief executive Sam Altman had also discussed pacing development with White House officials. Gates’ memo raises a separate concern: a preliminary survey associated heavier AI use with reduced critical-thinking ability, particularly among younger people.

The article therefore argues that even beneficially aligned AI could weaken human reasoning, although it provides no detailed methodology for the survey or specific mitigation measures.