Microsoft AI CEO Mustafa Suleyman is calling for direct government involvement in evaluating AI models before public release, arguing that the industry should not set its own capability limits. Microsoft has signed formal agreements with the US Center for AI Standards and Innovation and the UK’s AI Security Institute for pre-release testing of its frontier AI models. Microsoft President Brad Smith reinforced this message the day before, advocating for independent AI safety evaluators, suggesting a coordinated strategy from the company’s leadership. Suleyman has been pushing for strict red lines on AI capabilities, particularly those posing existential-scale risks. Questions remain about whether government agencies have the technical capacity to meaningfully evaluate cutting-edge AI systems, as these bodies are relatively new.
Source: Read the original article
