Particle.news
Download on the App Store

Anthropic Safeguards Lead Resigns, Warning ‘The World Is in Peril’

His exit intensifies scrutiny of Anthropic's safety posture.

Overview

  • Mrinank Sharma posted his resignation letter on Feb. 9, saying he will pursue a poetry degree, focus on writing, and move back to the UK.
  • Sharma led research on AI safeguards, including generative-AI sycophancy, AI-enabled biothreat risks, and ways assistants could erode human-centered behavior.
  • He wrote that it is difficult for organizations to let stated values govern actions, describing pressures to set aside what matters most.
  • The departure tracks with wider industry unease, including an OpenAI researcher leaving over ads in ChatGPT as Anthropic ran commercials criticizing that decision.
  • Anthropic promotes a safety-first mission as a public-benefit corporation but continues to face credibility questions after a $1.5bn 2025 settlement with authors over training data.