Anthropic safety chief resigns as company pursues massive 350 billion dollar valuation

The self-proclaimed "safety-first" AI laboratory Anthropic is facing an internal identity crisis as it seeks a record-shattering $350 billion valuation that would crown it the most valuable AI startup in history. Just days after the release of Claude Opus 4.6, a model designed to dominate the enterprise market, Mrinank Sharma, the head of the company’s Safeguards Research team, announced his resignation in a move that has sent shockwaves through the tech community. This departure signals a growing rift between the company’s original mission to build responsible AI and the immense commercial pressure of a $20 billion fundraising round.

In a cryptic note posted to X, Sharma warned that the world is "in peril" from a series of interconnected crises. Referencing the poet Rainer Maria Rilke, Sharma wrote that the industry is nearing a threshold where "our wisdom must match our power to shape the world." More damningly, he admitted that the relentless pace of development often forced values to be "set aside" within the organization.

The timing of the exit is a major headache for CEO Dario Amodei. Anthropic has seen its run-rate revenue jump from $1 billion to over $5 billion in just six months, fueled by aggressive product launches. However, this growth has come at a cost. Sharma’s final project reportedly focused on how AI assistants could "distort human behavior," suggesting that the "agentic" capabilities of Claude 4.6 may have crossed ethical lines the safety team was not yet ready to endorse.

This isn't the first high-profile departure; last week, leading scientists Behnam Neyshabur and Harsh Mehta also left the firm. Investors, including Blackstone and MGX, are pouring capital into Anthropic as it targets a valuation nearly double that of OpenAI’s last round. Yet, if the company’s internal safety apparatus is crumbling, that $350 billion price tag may come with unmanageable liabilities. Anthropic was founded by OpenAI defectors who left precisely because they feared commercial interests were eclipsing safety; today, the student appears to be following the teacher’s controversial playbook.

For now, Sharma plans to pivot to poetry, leaving behind a company that is technically brilliant but ethically frayed. As Anthropic prepares to finalize its massive capital injection, the question remains whether it can maintain its soul while chasing a valuation that demands nothing less than global dominance.