Anthropic's senior AI safety researcher Mrinank Sharma quits, says 'the world is in peril'

Mrinank Sharma, a senior AI safety researcher and head of the Safeguards Research Team at Anthropic, has resigned from the company, marking a rare and public departure from one of the world’s most influential artificial intelligence firms.

Sharma, who joined Anthropic in 2023, announced his decision through a public resignation note in which he framed his exit not as a career move, but as a moral and philosophical choice. Anthropic, led by chief executive officer Dario Amodei, is widely known for its Claude AI models and its emphasis on building safer and more responsible artificial intelligence systems.

Sharma’s resignation comes at a time when debates around AI risk, alignment and the societal impact of advanced models are intensifying globally.

‘The world is in peril’

In his resignation letter, Sharma warned that the dangers facing humanity go beyond artificial intelligence alone.

“The world is in peril," he wrote, adding that the threat does not come only from AI but from “a whole series of interconnected crises unfolding in this very moment."

He suggested that humanity’s ability to shape and influence the world has grown much faster than its collective wisdom, creating a dangerous imbalance that could have serious consequences.

Sharma also pointed to a gap between stated values and real-world action inside large organisations.

“Throughout my time here, I’ve repeatedly seen how hard it is to truly let our values govern our actions," he wrote, referring both to himself and to the broader institutional environment.

Turning away from AI work

Rather than continuing his work on improving AI systems, Sharma said he feels drawn in a different direction.

He noted that while efforts to make AI more transparent or less sycophantic are important, they no longer feel like the most urgent response to the current moment. Instead, he said he wants to focus on writing as a way of engaging with the world more directly.

Sharma argued for the importance of “poetic truth alongside scientific truth," saying both are necessary to understand reality during times of rapid technological and social change.

He added that he now hopes to pursue a poetry degree and devote himself to what he described as “the practice of courageous speech."

Who is Mrinank Sharma

Mrinank Sharma is an AI researcher of Indian origin with an academic background from leading global institutions. He holds a DPhil in Machine Learning from the University of Oxford and a Master of Engineering in Machine Learning from the University of Cambridge.

At Anthropic, Sharma led the Safeguards Research Team, which was set up in February last year. The team focused on reducing risks in advanced AI systems, including behavioural concerns and alignment challenges, as the company positioned itself as a leader in AI safety.

His departure stands out in an industry where senior researchers rarely leave on openly philosophical grounds, and it adds to broader questions about whether technical safety work alone is enough to address the challenges posed by rapidly advancing artificial intelligence.