Anthropic scientist exits over AI safety concerns, warns of existential risks

A scientist at Anthropic, the artificial intelligence research firm founded by former OpenAI executives, has resigned and publicly warned that the company and broader industry may not adequately prepare for existential risks posed by advanced AI systems. The departure marks another instance of technical talent leaving leading AI organisations over safety and alignment concerns.

The departing researcher's warning reflects escalating tensions within the AI industry regarding the pace of development versus the maturation of safety protocols. Anthropic was established specifically to focus on AI safety, positioning itself as more cautious than competitors, yet the resignation suggests even those standards may be insufficient in the view of some team members.

The exodus from major AI labs comes as the sector experiences unprecedented investment and rapid capability advancement. Companies including OpenAI, Google DeepMind, and others have seen researchers depart over disagreements about safety priorities, with some choosing to establish independent research groups or shift to policy roles.

Anthropically has emphasised its commitment to responsible AI development through techniques like constitutional AI and extensive red-teaming. However, the latest departure underscores persistent anxiety among AI researchers about whether current safeguards can address risks as systems become more powerful and autonomous.

The incident adds pressure on governments and regulators to establish clearer frameworks for AI oversight, even as industry leaders maintain that internal safety measures are advancing alongside capability gains.