EU votes to proceed with AI Act which could ban much facial recognition usage
For many use cases facial recognition will be categorised 'unacceptable', along with social scoring and cognitive behavioural manipulation, and outlawed, as per the current draft
The European Parliament today agreed to finalise the AI Act, beginning a process that will result in the world's first law specifically related to AI.
In the AI Act, The EU's approach diverges from that of other jurisdictions in that it proposes grouping AI solutions according to the potential risk they pose, with four categories: minimal, limited, high and unacceptable. The category into which a particular solution falls will decide which governance rules apply.
A minimal-risk application might include a grammar checker or music recommender, whereas a system that approves a bank loan based on credit scores and other characteristics of the borrower would be high risk, as bias within the system could lead to damaging consequences for the citizen.
The "unacceptable" category contains systems considered a threat to individuals or groups, including cognitive behavioural manipulation of people or specific vulnerable groups, social scoring, and remote biometric identification systems, such as facial recognition. In most situations, these will be banned.
In a BBC interview on Wednesday, EU competition commissioner Margrethe Vestager said that rather than the much-publicised fears of AI leading to human extinction, she was most worried about the risks in the here-and-now, including the potential for AI tools to amplify bias, increase discrimination and spread disinformation, and that criminals will pick up on their use faster than law enforcement.
"If your social feed can be scanned to get a thorough profile of you, the risk of being manipulated is just enormous, and if we end up in a situation where we believe nothing, then we have undermined our society completely," she said
However, she dismissed calls to halt the development of AI as impractical.
On the subject of facial recognition technology, which falls into the "unacceptable" category, she said. "We want to put in strict guardrails so that it's not used in real-time, but only in specific circumstances where you're looking for a missing child or there's a terrorist fleeing."
The Act is expected to become law by 2025. Before that it needs to be voted on by the European Commission, Parliament and Council. Under the current wording, violators could see fines of €40 million, or 7% of worldwide turnover.
Rest of the world
In contrast to the EU, the UK has proposed a more light-touch business-friendly approach. While the government has pivoted from being rather dismissive of AI safety to wanting to be the "geographical home of AI regulation", it has yet to publish any detailed regulations or guidance, information about compliance or coherent strategy. In April, it announced a £100 million Foundation Model Taskforce to promote AI innovation in the UK, into which AI safety was also rolled.
See also: AI giants agree to open up their models to UK government for research and safety
Similarly, the US is battling to balance AI safety with maintaining a global competitive advantage, particularly over China. It has issued a hodgepodge of executive orders, and a high-=level blueprint for an AI Bill of Rights, but so far there has been little clarity or agreement.
Meanwhile, UN secretary-general António Guterres this week called for a global AI regulator along the lines of international bodies that govern nuclear technology.
The draft Code of Conduct for Information Integrity on Digital Platforms, would require governments and tech companies to agree to a series of standards and protocols around data access, privacy, disinformation and hate speech.
"New technology is moving at warp speed, and so are the threats that come with it," he said. "Alarm bells over the latest form of artificial intelligence (AI) — generative AI — are deafening, and they are loudest from the developers who designed it."