What is the EU agreement and why is it significant?
The EU on Friday came to a political agreement on the rules for governing AI, bringing it closer to being the first major global power to establish enforceable AI legislation.
European Union president Ursula von der Leyen in a statement called it a ‘historic moment’.
“Today’s agreement focuses regulation on identifiable risks, provides legal certainty and opens the way for innovation in trustworthy AI. By guaranteeing the safety and fundamental rights of people and businesses, the Act will support the human-centric, transparent and responsible development, deployment and take-up of AI in the EU,” she said.
First proposed in 2021, the EU AI Act had classified AI systems according to their level of risk to cause harm, categorising banned practices, high-risk systems and other AI systems.
Discover the stories of your interest
What does the agreement lay down?
The agreement sets obligations on general purpose AI systems and foundation models with regard to ensuring technical documentation, copyright law compliance and transparency around content used for training models before their market release, according to Reuters.Models with systemic risk would also have to take steps to evaluate and mitigate them, conduct adversarial testing, and report serious incidents as well as their energy efficiency to the European Commission.
The deal further bans cognitive behavioural manipulation, indiscriminate scraping of facial images from the web or from CCTV footage, social scoring, and biometric identification to profile people by their race or sexual orientation, or by their political, religious and philosophical beliefs.
For governments, the pact limits the use of AI for real-time biometric surveillance to public spaces, for victims of specific crimes, searches for suspects of the most serious crimes, and prevention of threats such as terrorist attacks.
What were the contentious issues?
The two most contentious issues were the regulation of foundation models and the use of AI for biometric surveillance.
Some of the largest members – France, Germany and Italy – had sought to limit regulation of generative AI models, instead pushing for self-regulation by companies.
Similarly, governments of member countries were in favour of AI in biometric surveillance where relevant for national security, defence and military uses, according to Reuters. The EU legislators wanted to ban this.
This deal is provisional. The details for the final deal are to be hammered out in the coming days.
The European Parliament will vote on the AI Act proposals early next year, but legislation would not take effect until 2025, according to the BBC.
Von der Leyen said until the law is fully in force, “we will support businesses and developers to anticipate the new rules. Around 100 companies have already expressed their interest to join our AI Pact, by which they would commit on a voluntary basis to implement key obligations of the Act ahead of the legal deadline.”
The EU rules are expected to serve as a blueprint for other countries mulling AI regulation.
“Our AI Act will also make a substantial contribution to the development of global guardrails for trustworthy AI,” Von der Leyen said. “We will continue our work at international level, in the G7, the OECD, the Council of Europe, the G20 and the UN.”