In what may be one of the first such attempts in the world, European Union lawmakers agreed earlier this month on the text of a law that aims to classify AI risk and regulate its use.

Called the AI Act, the law – which still must pass a vote in the EU Parliament – aims to create rules governing various uses of artificial intelligence, along with classifying the risk presented by the technology in various use cases: from unacceptable (and therefore illegal) to high, limited, and minimal risk.

Uses deemed unacceptable by lawmakers include so-called “social scoring systems,” similar to those already being used in countries such as China, along with gathering facial images from the internet or CCTV footage and emotion recognition technology in workplaces and schools.

One thing the law does not do is lay out any technical specifications, with the lawmakers instead choosing to allow the industry to define those themselves. Instead, the law aims to be a “high level” set of guidelines that can more easily adapt to new and changing uses of this developing technology.

While the EU’s AI Act, which will not go into effect until 2025 at the earliest, is among the first of its kind in the world it is likely to not be the last. Lawmakers in other nations such as the United States are undoubtedly watching the development of the EU’s laws and could even use it as the basis for their own proposals.

AI Risk has been something on a lot of peoples’ minds, including our own, so we have finished the first iteration of our very own Artificial Intelligence Policy boilerplate that is available to download to help!