Former Anthropic Safeguards Lead Resigns Over AI Perils, Bioweapons Concerns

Mrinank Sharma, who formerly led the safeguards research team at Anthropic, announced his departure from the company in a letter to colleagues on Monday.

Sharma cited his concerns regarding artificial intelligence (AI) perils, bioweapons, and a series of interconnected global crises as reasons for his resignation.

Achievements During Tenure

During his tenure, Sharma stated he achieved his objectives. These included developing defenses against AI-assisted bioterrorism and conducting research on AI's effects on humanity. He noted the difficulty in consistently aligning values with actions within organizations and broader society.

Sharma's Future Endeavors

Sharma's future plans involve pursuing work that aligns with his integrity, exploring a degree in poetry, and dedicating himself to courageous speech.

Broader Context: Recent Departures from Anthropic

Sharma's departure is one of several recent exits from Anthropic. Other employees like Harsh Mehta and Behnam Neyshabur have also left. Former AI safety researcher Dylan Scandinaro, for instance, has joined OpenAI.

Anthropic's Recent Organizational and Funding Developments

In recent organizational developments, Anthropic has announced new hires, including CTO Rahul Patil. The company is reportedly in discussions to raise new funding, potentially valuing it at $350 billion. Anthropic also recently introduced an upgraded AI model, Claude Opus 4.6.