The European Parliament and European Council reached a deal Friday on AI regulation to protect consumers from high-risk uses of the technology while promoting innovation in a controlled environment.
The deal could become the international benchmark on AI. 
The EU AI Act, proposed in June, sets the ground rules for uses of general-purpose AI, according to today’s Parliament release.
“The AI Act emphasizes consumer protection, aiming to prevent AI and [generative AI] misuse and ensure transparency and ethical AI use,” Agustin Rubini, director analyst at consulting firm Gartner, told Bank Automation News. “This is expressed in its risk-level categorization of AI systems and compliance requirements for high-risk systems.”
According to the Parliament release, the proposal bans the following AI uses:
- Biometric categorization systems using sensitive characteristics including political, philosophical belief, sexual orientation or race;
- Use of untargeted scraping of facial images;
- Emotion recognition;
- Social scoring;
- Manipulation of human behavior; and
- Exploitation of vulnerability based on age, disability, social or economic situation.
AI at FIs
With safeguards in mind, regulators did not want to halt AI innovation, according to the release.
“The EU AI Act is set to bring significant regulations to the financial services sector,” Rubini said. “It categorizes AI systems by risk, with high-risk areas like credit scoring facing strict oversight. This could mean increased compliance costs, but the act aims for a balance that doesn’t overly hinder innovation.”
In fact, the regulation pushes for real-world testing and regulatory sandboxes to develop AI before it hits the market, according to the release.
Enforcing the act
Parliament’s internal market and civil committee must vote on the proposal, which is expected when it meets next Jan. 24-25, according to the release.
After the act is implemented, non-compliance would lead to fines beginning at 35 million euros ($37.7 million), according to the release.
“The act has the potential to become a global benchmark for AI regulation,” Rubini said.
“It is comprehensive in scope, has [a] strict penalty regime for non-compliance and has a broad definition of AI,” Rubini said.
Get ready for the Bank Automation Summit U.S. 2024 in Nashville on March 18-19! Discover the latest advancements in AI and automation in banking. Register now.
