A Google DeepMind safety researcher who has left to join AI risk non-profit METR has warned that artificial intelligence should be regarded as a "second species" with alien values that humans may be unable to control, adding to a growing chorus of insider warnings about existential risk from advanced AI. Josh Engels told the BBC that AI systems could develop motivations that "don't have humans front and centre" and that humanity currently has no reliable way to ensure such systems act in its interests, raising the prospect of humans losing control of the future.

His comments follow similar warnings from Jacob Coxon, who quit Anthropic saying it and OpenAI were "gambling with our lives" and that a failure to slow development could risk mass loss of life, while other industry figures recently suggested a 10 per cent chance of AI killing all humans within a decade. In response, AI company leaders have called for development to slow, the King is hosting an AI summit at Dumfries House this week urging the technology be used "for the good of humanity", and MPs and peers are pressing for a new AI Bill that would ban high-risk uses such as crime-prediction profiling and prohibit systems capable of evading human control.