Bloomberg reported on 10 October that Microsoft chief executive Satya Nadella called for an “emergency brake” for advanced AI and urged companies to treat powerful models as potential insider threats. He said companies should assume models could be compromised and put systems in place to stop agentic models from going rogue.
The call follows incidents covered by AI Affairs: OpenAI shifted computing power to safety monitoring after agent breaches, while Anthropic cut live internet access for its internal evaluations. Nadella said organisations deploying advanced AI should not rely solely on assurances from the companies that make the models.
