AI firms lack control over models, warns industry insider

Jacob Coxon, who previously worked at prominent AI developers OpenAI and Anthropic, has told legislators that major firms in the sector are failing to adequately manage the systems they create. Coxon's comments underscore growing concerns among regulators and industry observers about the governance and oversight of increasingly powerful AI models. The former employee characterised the approach taken by leading companies as extremely reckless given the potential consequences of systems that operate beyond their creators' full understanding or control. His testimony adds to a mounting body of evidence suggesting that rapid commercial development is outpacing the implementation of robust safety measures. Regulatory bodies across multiple jurisdictions have begun scrutinising AI development practices more closely, with particular focus on whether companies possess sufficient safeguards to prevent unintended consequences. The concerns raised highlight the tension between competitive pressure to deploy advanced AI systems and the need for comprehensive testing and control mechanisms before widespread deployment.