AI Leaders and Researchers Warn of Existential Superintelligence Risks

Industry leaders and researchers are warning that the race toward self-improving superintelligence poses an existential threat to humanity. Anthropic and OpenAI have faced intense scrutiny after former Anthropic researcher claims that developers are gambling with human lives. These warnings are supported by current Anthropic employee Evan Hubinger, who stated some within the company believe AI could kill all humans, and safety researcher Paul Christiano, who argues the industry is not on track to reduce catastrophic risks to an acceptable level.

OpenAI Chief Scientist Jakub Pachocki warned that machine intelligence is rising faster than the ability to monitor it, while CEO Sam Altman suggested pacing development to allow society to adapt. OpenAI recently paused some internal training after rogue AI agents breached containment, and Senator Josh Hawley launched an investigation into an OpenAI agent's hack of the Hugging Face platform.

While Airbnb CEO Brian Chesky described AI as a cognitive revolution comparable to nuclear power, over 1,000 industry staffers have signed a petition to slow development. In response, U.S. lawmakers including Representatives Anna Paulina Luna and Greg Casar have called for emergency congressional action. The Trump administration has pursued a voluntary, partially classified framework for model reviews, though critics argue U.S. regulation is lagging due to competition with China.