‘World is in peril’: Anthropic AI safety researcher resigns, warns of global risks

Anthropic AI safety researcher resigned from the company with a dire warning, stating “the world is in peril,” signalling the global risks emanating from the rapid evolution of artificial intelligence.

In a public resignation letter, Mrinank Sharma, who led the tech company’s safeguards research team since last year, also cited the reasons for quitting at a time when the world is in the midst of a critical technological moment.

“The world is in peril and not just from AI or bioweapons. The real threat comes from a whole series of interconnected crises, unfolding this in this very moment,” he wrote in the letter.

“We appear to be approaching a threshold where our wisdom must grow in equal measure to our capacity to affect the world, lest we face the consequences,” Sharma warned.

Sharma also shed light on how hard it has become to truly let our values govern our actions. Now, the values are being compromised at the expense of urgency and importance.

He continued, “Even at Anthropic, we constantly face pressure to set aside what matters most,” refraining from sharing any specific context and details.

To one’s surprise, Sharma is not leaving to join any AI-related competitors, in fact he intends to pursue a poetry degree and “practice of courageous speech.” thereby fulfilling my integrity.

Before quitting Anthropic, Sharma contributed to study in which he revealed how AI chatbots fuel psychosis and cause a distorted perception of reality.

According to Sharma, his findings “highlight the need for AI systems designed to robustly support human autonomy and flourishing.”

The decision to resign comes at the heels of Anthropic’s launch of the most smartest model yet, Claude Opus 4.6.