"We can’t have non-state actors or China or folks who we don’t want accessing our cloud to train their models" — US Government wants to bring in tough new regulations to keep tabs on AI

News
By Benedict Collins
published

Greater regulation for AI safety

IT
(Image credit: Shutterstock / carlos castilla)

The Biden administration is set to use federal law to levy new rules on companies that use or train artificial intelligence (AI) tools.

The rule, piggybacking on the existing Defense Production Act, will mean that companies will have to tell the government when they are training new AI models.

Another policy that could potentially be introduced would require cloud companies to monitor and, if necessary, ban foreign entities from training their AI models on US data.

Military priority for AI?

The Defense Production Act was signed into federal law during the Korean War to prioritize US industry towards the war effort. Part of ensuring the safety of AI includes protecting AI models from potential poisoning and theft by foreign powers such as China and Russia.

The new rule would allow government access to AI pioneers such as Google and Amazon to ensure that current and future rules on AI safety are enforced. The Biden administration is also seeking to regulate access to cloud computing companies by vetting the identities of cloud service customers to prevent malicious foreign access.

According to Reuters, Commerce Secretary Gina Raimondo said “We can’t have non-state actors or China or folks who we don’t want accessing our cloud to train their models.” 

These latest AI regulations follow an executive order set out last year that “establishes new standards for AI safety and security, protects Americans’ privacy, advances equity and civil rights, stands up for consumers and workers, promotes innovation and competition, advances American leadership around the world, and more.”

Via SiliconAngle

More from TechRadar Pro

Benedict Collins
Benedict Collins
Staff Writer (Security)

Benedict Collins is a Staff Writer at TechRadar Pro covering privacy and security. Benedict is mainly focused on security issues such as phishing, malware, and cyber criminal activity, but also likes to draw on his knowledge of geopolitics and international relations to understand the motivations and consequences of state-sponsored cyber attacks. Benedict has a MA in Security, Intelligence and Diplomacy, alongside a BA in Politics with Journalism, both from the University of Buckingham.