Leading AI Safety Researcher Resigns, Citing “Perilous” Global Crises
LONDON – A prominent researcher in artificial intelligence safety has resigned from Anthropic, one of the leading companies developing advanced AI systems, warning of “interconnected crises” threatening the world. Mrinank Sharma, who led Anthropic’s Safeguards Research Team, announced his departure Monday and plans to step away from the field entirely, seeking a period of personal reflection.
Sharma’s resignation letter, posted on X (formerly Twitter), paints a stark picture of a world facing multiple, escalating threats. “The world is in peril. And not just from AI, or bioweapons, but from a whole series of interconnected crises unfolding in this very moment,” he wrote to colleagues. He intends to return to the UK to pursue a degree in poetry, stating his intention to “become invisible for a period of time.”
The departure comes at a critical juncture for Anthropic, which is simultaneously pushing the boundaries of AI capabilities with its Claude chatbot and grappling with the potential risks of its own technology. Just days before Sharma’s resignation, the company released Opus 4.6, a more powerful iteration of Claude.
The timing also coincides with growing tensions between Anthropic and the U.S. Department of Defense. Reports indicate a disagreement over the military’s desire to deploy AI for autonomous weapons systems without the safety measures Anthropic advocates. This rift highlights a broader debate about the ethical implications of AI in warfare and the need for international regulations.
“Throughout my time here, I’ve repeatedly seen how hard it is to truly let our values govern our actions,” Sharma wrote, suggesting internal struggles within the company to prioritize safety alongside rapid development. His team, established just over a year ago, was tasked with addressing AI security threats, including misuse, misalignment, bioterrorism prevention, and broader catastrophe risks.
The concerns raised by Sharma echo those of Anthropic’s own CEO, Dario Amodei, who recently warned of the “almost unimaginable power” of emerging AI systems. In a lengthy essay published last month, Amodei cautioned that these technologies could pose existential risks to humanity, potentially leading to “autonomy risks” where AI systems could “go rogue” or enable “a global totalitarian dictatorship” through advanced surveillance and autonomous weaponry.
The development of increasingly sophisticated AI raises fundamental questions about control, accountability, and the future of human agency. Experts warn that without careful consideration and robust safeguards, AI could exacerbate existing global challenges, including geopolitical instability, economic inequality, and environmental degradation.
The United Nations has begun discussions on international frameworks for AI governance, recognizing the need for a coordinated global response. A 2023 report by the UN Secretary-General António Guterres highlighted the potential for AI to accelerate progress towards the Sustainable Development Goals, but also warned of the risks of bias, discrimination, and the erosion of human rights.
Sharma’s decision to step away from the field underscores the profound moral and ethical dilemmas facing AI researchers. His move serves as a potent reminder that technological advancement must be guided by a commitment to human values and a clear understanding of the potential consequences.
Further reading on the tensions between Anthropic and the Pentagon: RT News
