OpenAI and Anthropic back binding AI safety rules at UK hearing

OpenAI and Anthropic representatives backed a statutory AI safety framework at a British parliamentary hearing on Tuesday, WIRED reported. Anthropic called for binding rules, while OpenAI supported mandatory elements beginning with incident reporting and an international convention. Google DeepMind and Meta declined to endorse new legislation. Officials also confirmed that Britain's AI Security Institute relies on companies and foreign governments for access to frontier models. The institute cannot prevent a company from releasing a model it considers dangerous. Anthropic said the institute now has access to Mythos 5.1, after previously being unable to provide full unrestricted access because of US government constraints.