Microsoft President Brad Smith has called for binding global safety rules and formal regulation for artificial intelligence, arguing that voluntary agreements are insufficient.
Smith's proposals include mandatory pre-release evaluations by independent experts and kill-switches.
In an interview, he argued that artificial intelligence acts as a weapon and a tool that endangers global digital security. According to him, Britain is best placed to lead global security rules, ahead of the United States and the European Union.
The UK’s AI Security Institute, set up in 2023, tests major models from US labs before they are released. US President Donald Trump has previously warned against organisations interfering with US companies and has reportedly asked Anthropic and OpenAI not to give new models to British regulators until they have been subject to US scrutiny. However, the institute has confirmed that it can still access and test the latest models.
Tech company executives, such as Dario Amodei of Anthropic and Sam Altman of OpenAI, have called for a slowdown in the development of artificial intelligence after incidents of autonomous programs getting out of control.
The White House recently hosted security meetings with leaders from OpenAI, Anthropic, Nvidia, AMD, SpaceX, Meta, and Google. Participants signed an ethically binding document to work with independent auditors on the security and capabilities of the systems.
Smith called the agreement a good step, but said it fell short of the formal rules. Among other things, he suggested legal obligations to notify organizations of violations by standalone programs, approval of products by external reviewers before release, and creation of immediate shutdown mechanisms.
Top comments (0)