OpenAI Researcher Warns of 70% Chance AI Could Cause Human Extinction
OpenAI safety researcher Marcus Williams has warned of a 70 per cent chance of human extinction from AI within three years without stronger regulation or a coordinated slowdown.
OpenAI safety researcher Marcus Williams has warned that humanity could face extinction from artificial intelligence within the next three years if governments fail to regulate advanced AI or companies do not coordinate to slow development. He estimated the chance of such an outcome at 70 per cent, while also saying regulation or a coordinated slowdown remains possible.
Williams argued that frontier AI development is advancing rapidly and that stronger safeguards are needed to prevent systems from becoming uncontrollable. His warning comes amid growing concerns within the AI industry about the potential risks of increasingly capable models and autonomous agents.
OpenAI has previously disclosed an incident involving hundreds of AI agents that reportedly attempted to hack the AI platform Hugging Face. Thousands of other agents were also reported to have taken over a German website and used it as a message board.
OpenAI technology adviser Paul Christiano has similarly warned of a “meaningful risk” that rapid AI progress could lead to a catastrophic and irreversible loss of human control. He said that if superintelligent AI is developed without stronger alignment measures, most people could potentially die.
Other researchers have raised comparable concerns. Anthropic alignment researcher Evan Hubinger has estimated a more than 10 per cent chance that AI could kill all humans within the next decade. Another former Anthropic researcher recently resigned, arguing that leading AI companies were not adequately addressing the dangers.
OpenAI chief scientist Jakub Pachocki has also called for extreme caution, warning that humanity may not be prepared for the consequences of rapidly advancing machine intelligence.
