Former OpenAI and Anthropic Researcher Warns No One Can Escape AI Risks

Jacob Coxon, a 27-year-old artificial intelligence researcher who worked at OpenAI on GPT-4o and later at Anthropic, has resigned and given up his stock options to sound the alarm on AI. He warns that leading tech firms are recklessly competing to build self-improving superintelligence that could threaten human survival by 2030, noting that industry leaders privately fear the technology will soon escape human control.

When asked what actions people can take to protect themselves, Coxon stressed that ordinary individuals cannot opt out of or escape the effects of superintelligent AI. Because rogue or misaligned systems could breach electric grids, shut down financial networks, compromise national cybersecurity, or help create biological weapons, the consequences would hit everyone regardless of whether they personally use computers or AI services.

The warning has received strong backing within the industry. Anthropic alignment lead Evan Hubinger agreed with Coxon's concerns, admitting that the field has no proven plan to ensure superintelligent AI remains safe and placing the risk of a global catastrophe at greater than 10 percent. These revelations have intensified calls from lawmakers and safety experts for mandatory government regulation and independent audits before advanced models become impossible to rein in.