Anthropic AI safety researcher quits to study poetry warns 'world is in peril'

Mrinank Sharma senior AI safety researcher at Anthropic has quit his position citing cryptic reasons that “the world is at peril”. His resignation letter was philosophical and poetic, a blend of technical concerns and humanities existential crisis. He warned that the world faces a series of interconnected threats from ai to bioweapon, to societal and structural challenges that far outpaces the wisdom of humanity. The resignation sound vey little like a corporate memo with citations and references to Rainer Maria Rilke and William Stafford, and which expressed interest in “poetic truth alongside scientific truth."

His resignation letter: “The world is in peril. And not just from AI, or bioweapons, but from a whole series of interconnected crises.” He further added “repeatedly seen how hard it is to truly let our values govern our actions”. The Anthropic AI is a startup that is known for its Claude AI models. It has a built-in ethical guidance that makes it appealing for sensitive applications. The company was founded on the idea of "safe" AI lab. It processed itself to be a "public benefit corporation dedicated to securing AI's benefits and mitigating its risks".

Mrinank had been engaged in safety work like studying AI sycophancy and defenses against AI-assisted bioterrorism.

According to Mrinank personal and institutional values were hard to uphold in practice. He wants to move back to UK and study poetry and engage in writing coaching, community building, and contemplation, a shift away from tech toward cultural and philosophical engagement.

Anthropic has recently been under scrutiny for unethically plagiarising contents of researchers to train it's AI models. It agreed to settle the class action lawsuit by paying $1.5 billion. Recently Anthropic has released a commercial critical of Opena AIs move to start running ads in the platform.