Advertisement

AI Act to arrive soon: Next month, EU parliament will officially vote

AI Act to arrive soon: Next month, EU parliament will officially vote
Judges wooden gavel with EU flag in the background. Shutterstock / Marian Weyo

The EU AI Act has been subject to proposed changes by MEPs, including tighter regulations on the use of biometric facial scanning in public areas.

The long-awaited AI Act from the European Commission is getting closer to being fully implemented.

A draft of the negotiating mandate AI regulations was approved by the internal markets and civil liberties committees with 84 votes in favour, 7 against, and 12 abstentions.





The laws will be adopted if the entire European Parliament approves the draft negotiating mandate. Next month's final vote is anticipated.

The AI Act was promoted as a mechanism to limit what the EU considered to be "high-risk" AI activity as early as 2021. In the same year, a proposal for the AI Act was made. The EU's main concern is protecting citizens' rights as technology develops.

MEPs pushed for a number of revisions to the proposal that would make AI systems safe, transparent, non-discriminatory, and ecologically sustainable.

They also urged for a unified definition of AI that is pretty neutral so that it may be used to describe the technology as it develops.

A long time in the making

The Commission has already tried to include moral AI principles into its legislative process. For instance, it published the AI Liability Directive in September of last year as a means of defending consumers' legal rights against businesses that market goods and services with AI.

The antiquated EU consumer responsibility regulations did not address potential harm to users of AI technology, such as drones.

The Commission has also been asked to regulate AI in law enforcement, which makes use of tools like biometric facial recognition scans. A group of MEPs recommended a blanket prohibition on the use of this kind of technology by the government in November of last year out of concern for abuse.





Amendments

Today's amendments prohibited the use of AI technology, such as remote biometric identification systems, in public areas.

Both predictive policing AI technology and biometric categorization techniques that rely on delicate traits like gender and ethnicity will be outlawed.

The Commission decided to uphold the right to privacy and forbid the collection of biometric information from social media or CCTV footage for the purpose of building facial recognition databases.

Any use of AI that could be deemed "high-risk" would be strictly monitored. The disclosure that the content of a generative AI model, like ChatGPT, was produced by AI, is one of the transparency standards that must be met. A summary of any copyrighted data used for training will also need to be disclosed by companies adopting generative AI.

There will be some exceptions to the restrictions on AI, particularly when it comes to innovation.