Jacob Coxon resigned from Anthropic after three years doing pretraining research at both Open AI and Anthropic, publicly warning that both companies are racing toward self-improving superintelligence without adequate safety plans. He left before his equity vested, removing any financial incentive to stay silent.

Coxon, who joined Anthropic at 27, said engineers building AI earnestly believe it could kill all humans by the end of the decade. He does not call the company evil but accuses it of cutting corners under competitive pressure. Evan Hubinger, an alignment team lead at Anthropic, confirmed Coxon's assessment, estimating greater than 10% chance of catastrophic AI within a decade.

The article draws parallels to Frances Haugen, who resigned from Facebook in 2021 after uncovering internal research showing Instagram worsened body image issues for one in three teen girls. Both insiders chose to leave and warn the public rather than stay silent.

A 2026 survey of tech workers found 53% would steer newcomers away from the field, while 67% of developers spend more time debugging AI-generated code than writing it, reflecting a culture of exhaustion declining quality.