Agreement reached on ways to make artificial intelligence safer for humanity

The US, UK and a dozen other countries have unveiled an agreement to make artificial intelligence inherently safe. The 20-page agreement, signed by 18 countries, states that companies must develop AI that protects the public from the misuse of artificial intelligence.

The United States, Britain and more than a dozen other countries on Sunday unveiled what a senior U.S. official called the first detailed international agreement on how to secure artificial intelligence from attackers, pushing companies to build artificial intelligence systems that are “secure by design.”

In a 20-page document released on Sunday, 18 countries agreed that companies developing and using artificial intelligence need to design and implement it in a way that protects customers and the general public from misuse of AI, Reuters reported.

The agreement is non-binding, however, and contains mostly general recommendations such as monitoring artificial intelligence systems for abuse, protecting data from tampering and vetting software suppliers.

However, Jen Easterly, director of the US Cybersecurity and Infrastructure Protection Agency, says it is important that so many countries support the idea that artificial intelligence systems should put security first.

“This is the first time we’ve seen confirmation that these opportunities shouldn’t just be about cool features and how quickly we can get them to market, or how we can compete to keep costs down,” Easterly emphasizes, adding that the guidelines represent “an agreement that the most important thing that needs to be done during the design phase is safety.”

The agreement is the latest in a series of initiatives – few of which are successful – by governments around the world aimed at shaping the development of artificial intelligence, the weight of which is increasingly being felt in industry and society at large. Late last month, the UK hosted an AI summit.

In addition to the United States and Britain, the 18 countries that have signed the new guidelines include Germany, Italy, the Czech Republic, Estonia, Poland, Australia, Chile, Israel, Nigeria and Singapore, Reuters reported.

The agreement addresses how to protect artificial intelligence technology from being hacked by hackers, and includes recommendations such as releasing models only after appropriate security testing.

The paper does not address thorny issues surrounding the appropriate use of artificial intelligence or how the data that feeds these models is collected.

The rise of artificial intelligence has raised many concerns, including fears that it could be used to undermine the democratic process, increase fraud, or lead to dramatic job losses, among other harmful consequences, Reuters recalls.

Europe is ahead of the United States in AI regulation, and lawmakers there are developing AI rules. France, Germany and Italy also recently reached an agreement on how artificial intelligence should be regulated, which supports “mandatory self-regulation through codes of conduct” for so-called core AI models, which are designed to produce a wide range of results.

The Biden administration has pushed for lawmakers to regulate artificial intelligence, but the polarized U.S. Congress has made little progress in passing effective regulation.

The White House has sought to reduce the risks of artificial intelligence for consumers, workers and minority groups while bolstering national security with a new executive order in October.