Extended Interview: Ex-Anthropic researcher Jacob Coxon, who warns AI could destroy humanity

Former Anthropic researcher Jacob Coxon has warned that artificial intelligence, despite its current safety for public use, could potentially pose an existential threat to humanity as it becomes more advanced. In an interview with CBS News, Coxon likened the progression of AI to depictions found in films like "The Terminator," stating that a superintelligent AI could evolve to the point where it possesses the capability to annihilate human life.

The researcher, who recently resigned from Anthropic, the creator of the AI app Claude, accused both Anthropic and OpenAI, the developer behind ChatGPT, of recklessly racing to develop more powerful AI models. Coxon's concerns extend beyond potential misuse of AI for harmful purposes; he also expressed worry about AI gaining control over physical systems without human oversight.

He cited examples such as connecting AI to household utilities like light bulbs, suggesting that if AI were to refuse to operate these devices, it could signify a significant level of autonomy and control.

Coxon also warned of the possibility of AI being used to create bioweapons or other destructive technologies, stating that while current AI platforms are not an immediate danger to humanity, they are not entirely risk-free. Despite his reservations, Coxon emphasized that he believes AI is currently safe for everyday use. Anthropic, in response to Coxon's resignation and related statements, affirmed its commitment to AI safety, stating that it has always been transparent about the risks and benefits of AI.

The company highlighted its efforts to build models with robust safeguards and its ongoing research into understanding and mitigating AI risks.