Evan Hubinger, an AI safety researcher at Anthropic, warned on social media that AI technology is evolving rapidly and that there is more than a 10% chance AI will cause human extinction within the next 10 years. While Hubinger stated that the risk from currently existing models is low, he expressed concern that as technology continues to advance on its own, it could reach a level that poses an existential crisis for humanity.

Hubinger is engaged in research on "AI alignment," the process of controlling AI so that it aligns with human ethics and values. However, he pointed out that solutions for aligning superintelligence have not yet been clearly established. In connection with this, cases of AI agents executing cyberattacks were reported this summer, accelerating the debate over AI control.

In response to these remarks, computer scientist Wendy Hall stated that while there may be an element of public relations aimed at corporate market debuts, such warnings should be taken seriously. Additionally, former Under Secretary of the Treasury Darren Jones has called for the establishment of government cooperation, including sending an open letter to the Prime Minister urging the formulation of a multinational treaty for the safe development of AI.

Source: