In a stark warning delivered from the stage of a tech conference in San Francisco, Anthropic co-founder Daniela Amodei called for immediate, enforceable guardrails on artificial intelligence development. Speaking to a subdued audience of engineers, investors and policymakers, she argued that the current trajectory of AI research is unsustainable without a robust framework of human oversight. Her comments come as the industry races to deploy generative models across healthcare, finance and defence sectors, raising fears of unintended consequences.
Amodei, whose company positions itself as a safety-first alternative to mainstream AI labs, insisted that the technology must remain under human control at all decision points. She warned that without binding regulations, society risks ceding critical choices to opaque algorithms. Anthropic has long advocated for transparency in model training and usage, but Amodei's latest remarks reflect growing unease even among insiders.
She called for an international treaty akin to the Paris Agreement, but with teeth: mandatory audits, liability clauses and a veto power for governments over dangerous applications. Critics, however, question whether such measures would stifle innovation or merely push development underground. The speech sparked immediate reaction online, with some praising her courage and others accusing her of fearmongering.
For the average user, the debate may seem abstract, but Amodei's point is that AI already influences everything from loan approvals to criminal sentencing. As algorithms become intertwined with daily life, the question of who pulls the lever grows ever more urgent. The next twelve months, she argued, will determine whether we steer the technology or let it steer us.
