The European Union is considering an AI ban

Salesforce’s Vala Afshar said it best: “In the last 20 years, software ate the world. In the next 20 years, AI will eat software.”

Artificial Intelligence (AI) may well chomp away, but not without first being put on a diet.

The EU is considering a partial AI ban that will be formally announced April 21, according to a draft proposal picked up by Politico.

The proposed ban is the first of its kind

Europe’s legislative body will likely focus on “high risk” AI systems and could fine companies up to €20m or 4% of revenue if they don’t comply.

“High risk” refers to things like mass surveillance and social credit scores that can impact safety and privacy. AI for things like manufacturing and energy would likely be good to go.

The regulations include:

• A surveillance ban on AI systems that track people indiscriminately
• A ban on social credit scores that track individual behaviors, impact hiring and judiciary decisions, and rate trustworthiness
• Bias prevention measures like human oversight in testing datasets
• AI notifications that would be sent to people when interacting with AI systems

The catch: Experts say the rules are vague and leave room for loopholes.

The rules also come as AI investment in the US is hotter than ever:

• The National Security Commission on Artificial Intelligence recently called for $32B in annual nonmilitary federal spending on AI research
• Already in 2021, 442 US VC deals with AI startups were worth a combined $11.65B

US companies could likely be subject to the EU’s new rules. Though the draft could change come April 21, we have a feeling few companies will be big fans.