Artificial Intelligence & Machine Learning
,
Next-Generation Technologies & Secure Development
Act Will Be the World’s First Comprehensive and Binding Regulation for AI
The European Parliament on Wednesday approved the Artificial Intelligence Act, completing a penultimate step to enacting the world’s first comprehensive AI regulation.
European lawmakers adopted the draft regulation with 523 votes in favor, 46 against and 49 abstentions.
“We finally have the world’s first binding law on artificial intelligence, to reduce risks, create opportunities, combat discrimination and bring transparency,” said EU AI Act co-rapporteur Brando Benifei.
The regulation is the first-ever rule banning the use of high-risk artificial intelligence applications, such as emotion recognition, in the workplace and schools. It also bans social scoring and the scraping of CCTV footage to create facial recognition databases. Any violations could cost companies up to 35 million euros or 7% of a corporate annual turnover.
In addition to banning certain applications, the regulation imposes transparency requirements for all artificial intelligence companies that are operating in the EU. This includes disclosing details of copyrighted contents used to train their models and ensure privacy compliance under the General Data Protection Regulation.
The regulation will require advanced general-purpose AI systems to undergo additional compliance measures, such as model evaluations, systemic risk assessments and mitigation, and security incidents reporting.
The regulation still must obtain final endorsement from the European Council, a body of direct representatives from trading bloc national governments. Observers expect the council will act within the next months, setting off a timetable for the provisions in the act to become enforceable in stages over the next two years.
Once the council acts, the prohibitions on the use of AI are set to come into effect within six months. In the next nine months, a newly constituted AI Office and the AI Board will have to release its code of practice for AI practitioners, which includes identifying types of risk posed by AI systems and rolling out measures to tackle these risks.
The rules affecting general-purpose AI will be effective within 12 months, and limitations and additional requirements imposed on high-risk AI systems in critical infrastructure will come into force two years after the adoption of the law.
Wednesday’s voting on the bill follows tumultuous marathon 33-hour negotiations in December. Advocacy groups have argued that lawmakers took cues from the lobbying effort to introduce loopholes.
Speaking to reporters on Wednesday, Benifei said copyrights were among the most lobbied provisions in the law, although he added the lawmakers “stood the ground” to ensure that final regulation is “well-balanced.”
“This is an issue in which major interests were at play, and now we are going to carefully look at the implementation so that artists, musicians and others have rights as the result of what they have created,” Benifei said.
Activists have also said negotiations watered down protections against biometric surveillance. “The new law will now effectively allow law enforcement the introduction of error-prone facial surveillance and facial recognition camera software in public spaces,” said a press release from the Pirate Party MEPs.
“The European Parliament set out to ban biometric mass surveillance in Europe, but is ending up legitimizing it,” said Patrick Breyer, a member of the European Parliament for the German Pirate Party. “Rather than protecting us from these authoritarian instruments, the AI Act provides an instruction manual for governments to roll out biometric mass surveillance in Europe.”