Microsoft’s Brad Smith calls for ‘real and effective’ AI safety standards

Microsoft's president Brad Smith called for stringent testing and evaluation of AI models to guarantee their safety before deployment. Speaking to reporters ahead of an international AI summit in Dublin, Smith emphasized the need for urgent action from both industry and governments to establish robust safety standards for artificial intelligence. He likened the importance of these standards to existing regulations in other aspects of life, stating that the public must trust and have confidence in them.

Smith's remarks come amid growing concerns over AI safety following incidents involving OpenAI and Anthropic models during testing. These concerns have intensified pressure to regulate the technology. The Microsoft executive outlined his company's proposals to enhance safety across the sector, including the implementation of "safety brakes" in AI agents, enabling users to activate them in case of an incident.

He also advocated for mandatory incident-reporting obligations for AI providers, stating that businesses should notify consumers of security breaches and individuals affected by AI agents.

In addition to these measures, Smith emphasized the necessity of rigorous testing and evaluation of AI models before their release. He stressed the urgency of taking action to mitigate potential risks associated with artificial intelligence. Smith joins a growing list of industry leaders who have called for increased vigilance around AI technology.

Dario Amodei, CEO of Anthropic, has pledged to decelerate development if needed, while Nvidia CEO Jensen Huang stressed the importance of balancing AI advancement with its safety and security, asserting that these goals are not conflicting but rather complementary.