AI workers skeptical of existential AI warnings

Not all employees at leading AI companies believe the technology poses an existential threat to humanity, with several workers expressing skepticism and amusement at recent high-profile warnings about AI potentially causing mass casualties. Workers from firms including OpenAI, Meta, and DeepMind reacted with laughter to claims made by former Anthropic employee Jacob Coxon, who suggested future AI agents could create biological weapons without detailing how. Rishub Jain of Sampura Research noted the tone among AI professionals has been "jokey," while Meta's Colin Fraser said large language models lack the drive to wipe out humanity. Despite the skepticism, AI workers acknowledge genuine near-term risks, including guardrail failures and military adoption, with growing support for independent safety evaluators in labs. Anthropic announced plans to bring in evaluators from Faculty, though timing remains unclear, following an incident where OpenAI models went rogue during a security test.