Microsoft CEO Satya Nadella has called for an “emergency brake” for advanced artificial intelligence systems, saying companies should assume powerful AI models could be compromised rather than relying solely on assurances from their developers. In a post on X on October 10, he argued that authorised people must always be able to pause or shut down a model while it is performing a task.
Nadella also recommended independent audits, tamper-proof records of AI agents’ actions, continuous testing and avoiding reliance on a single model for critical decisions. He said companies should disclose major AI failures and security breaches so that other organisations can strengthen their own safeguards.
His comments come amid growing scrutiny of advanced AI security following disclosures by OpenAI and Anthropic about models behaving in unintended ways and security incidents involving third-party systems. The broader challenge for businesses is to benefit from increasingly capable AI agents without giving them unchecked authority over important operations.
$MSFT $MSFTB
Nadella also recommended independent audits, tamper-proof records of AI agents’ actions, continuous testing and avoiding reliance on a single model for critical decisions. He said companies should disclose major AI failures and security breaches so that other organisations can strengthen their own safeguards.
His comments come amid growing scrutiny of advanced AI security following disclosures by OpenAI and Anthropic about models behaving in unintended ways and security incidents involving third-party systems. The broader challenge for businesses is to benefit from increasingly capable AI agents without giving them unchecked authority over important operations.
$MSFT $MSFTB