Anthropic safety researcher warns AI carries over 10% chance of human extinction within a decade

A top safety researcher at AI firm Anthropic has sounded a severe alarm, estimating a greater than 10% chance that artificial intelligence could wipe out humanity within the next ten years. Evan Hubinger shared his stark assessment in a social media post, pointing to rapid technological advancements that could soon push systems toward posing an existential threat.

What People Are Searching For

Tech users and observers are searching for details behind the sudden wave of warnings coming directly from industry insiders. Following Hubinger's post and the departure of another researcher, Jacob Coxon, readers want to know what prompted such dire predictions about artificial intelligence and whether current safety guardrails are failing.

What Is Confirmed

Evan Hubinger stated on social media that while current risks from existing models remain low, he is worried about future systems developing the capability to improve themselves autonomously. His comments arrived alongside the resignation of Jacob Coxon, who argued that major AI labs are not acting responsibly as they race toward building superhuman systems. Independent computer scientist Dame Wendy Hall expressed shock at the public disclosures, questioning whether the disclosures might partly tie into upcoming market debuts, while noting investors should take the warnings seriously. Meanwhile, political figures have jumped into the debate, with US Senator Ted Cruz citing similar existential figures on his podcast and calling the warnings highly concerning.

What Is Unconfirmed

Specific details regarding how advanced models could physically result in human extinction have not been outlined by the researchers. Furthermore, reports indicating that Anthropic withheld its latest model from the UK's AI Security Institute remain unconfirmed by company officials, with Anthropic declining to comment on the situation.

The Bigger Picture

The disclosures highlight deep-seated anxieties within the AI alignment community, which attempts to build human ethical principles into autonomous software. Despite companies publishing safety reports indicating low risks of catastrophic harm, prominent figures across the sector continue to debate whether current alignment strategies are keeping pace with exponential technological acceleration.

Latest Status

Lawmakers and international advisors are stepping up calls for binding regulatory frameworks and multinational treaties to govern the development of superintelligence. As developers push boundaries, governments face mounting pressure to intervene before autonomous agents gain widespread operational power.

Frequently Asked Questions

• Who warned that AI could kill all humans?

  Evan Hubinger, a top safety researcher at Anthropic, published an assessment stating there is a greater than 10% chance AI could cause human extinction within a decade.

• Why did the Anthropic researchers speak out?

  The warnings emerged alongside the resignation of AI researcher Jacob Coxon, who stated that major labs are failing to act responsibly as they race toward superhuman systems.

• How have lawmakers reacted to the warnings?

  Lawmakers in the United States and abroad have expressed alarm, with officials calling for heightened caution, multinational safety treaties, and stricter government oversight.

• What is AI alignment?

  AI alignment is the field of research focused on building ethical principles and human values into artificial intelligence systems to ensure they remain safe and under control.