Anthropic's AI safety head steps down citing corporate pressure

A high-profile departure has rocked Anthropic: the company's AI safety lead Mriyank Sharma has stepped down, citing an internal values conflict. In a farewell letter posted on platform X that garnered millions of views, he thanked colleagues for their collaboration but emphasized that mounting economic pressure has led to decisions he can no longer support.

Sharma headed the Safeguards Research team and was considered a key expert in safe AI development. In his address, he stated that the world is in danger, describing the current situation as a polycrisis—a convergence of interconnected global problems. According to him, this involves not just risks from AI or biotechnology, but also growing geopolitical instability and systemic challenges that reinforce each other.

The letter particularly highlights the difficulty of maintaining principles in a corporate environment. Even companies that position themselves as safety- and values-oriented face market pressure and the need for compromises. Given that Anthropic has built its image as a more responsible alternative to other AI giants, Sharma's statement has resonated especially strongly.

Following his departure, he intends to completely change direction and pursue philosophy studies. The community has responded to his decision with respect, though some users believe he could have exerted more influence from within the company. Sharma himself notes that outside corporate constraints, he has more freedom to speak openly. Against the backdrop of aggressive promotion of the Claude chatbot and Anthropic's extensive advertising campaign, his resignation may affect how the company is perceived in the AI safety field.