Book a call

News · April 21, 2021

Brussels proposes risk-based rules for AI systems

// ai.news :: 2021

The European Commission has put forward the first broad legal framework for AI, sorting systems by risk level. Companies selling AI products into the EU would be covered wherever they are based.

What happened

  • The European Commission proposed on April 21, 2021 a regulation that groups AI systems into four levels: unacceptable risk (banned), high risk, limited risk and minimal risk, according to its press release.
  • Per the Commission, high-risk uses, including AI in employment, education and essential services, would need risk assessment, quality training data, activity logging, documentation, human oversight and high accuracy before going to market. Chatbots would have to tell users they are talking to a machine.
  • TechCrunch reported proposed fines of up to 6% of global annual turnover or 30 million euros for breaking the bans, and up to 4% or 20 million euros for breaching the high-risk rules.
  • TechCrunch also reported that the rules would apply to any company selling AI into the EU, and that adoption by the European Parliament and member states could take years.

Why it matters

Nothing changes this year: this is a proposal that still has to pass through the EU legislative process. But the direction is clear, and the requirements it lists (data quality, logging, documentation, human oversight) are good engineering practice anyway.

Teams building AI features should start keeping an inventory of where models are used, what data trains them and who can override their output. Most consumer and marketing uses look likely to fall into the lower-risk tiers, but features touching hiring, credit or education deserve a closer look early.

Sources

  • European Commission: Europe fit for the Digital Age: Commission proposes new rules and actions for excellence and trust in Artificial Intelligence
  • TechCrunch: Europe lays out plan for risk-based AI rules to boost trust and uptake

This post is CMM’s own summary and commentary on publicly reported news. Facts are attributed to the sources linked above; please read them for full details. CMM is not affiliated with these publishers or with the companies mentioned, and trademarks belong to their owners. Prepared with AI assistance. Corrections or requests: [email protected]. See our editorial policy.

Want help with this?

Tell us what you need, and we will suggest the right team model on a short call.

Talk to us