The European parliament is preparing tough new measures over the use of artificial intelligence, including forcing chatbot makers to reveal if they use copyrighted material, as the EU edges towards enacting the world’s most restrictive regime on the development of AI.
MEPs in Brussels are close to agreeing a set of proposals to form part of Europe’s Artificial Intelligence Act, a sweeping set of regulations on the use of AI, according to people familiar with the process.
Among the measures likely to be proposed by parliamentarians is for developers of products such as OpenAI’s ChatGPT to declare if copyrighted material is being used to train their AI models, a measure designed to allow content creators to demand payment. MEPs also want responsibility for misuse of AI programmes to lie with developers such as OpenAI, rather than smaller businesses using it.
One contentious proposal from MEPs is a ban on the use of facial recognition in public spaces under any circumstances. EU member states, under pressure from their local police forces, are expected to push back against a total ban on biometrics, said people with direct knowledge of the negotiations.
Agreement between MEPs, who have been fighting over measures to police artificial intelligence for close to two years, is critical to kick-starting broader negotiations over the AI Act. The proposed law would represent some of the toughest rules on the development of AI and comes in the wake of rising concerns about potential abuses of the technology.
Once parliament has agreed its position, member states, the European Commission and MEPs will together draft a final bill, aiming to pass the law before the end of the current European parliament term in 2024.
MEPs hope to finalise their position next week, Brando Benifei, one of the parliament’s leading negotiators, told the Financial Times.
Under current proposals for the AI Act, chatbots must tell users that they are not humans. Benifei said MEPs wanted further transparency.
“We want a list of the public disclosure of the material that is being used to train it because [authors] can go through other legislations and avenues to try to get paid for what is being used without their consent,” he said.
The proposal comes as visual media company Getty Images pursues a copyright case in the UK against Stability AI, claiming the maker of the free image-generating tool used millions of its copyrighted images without consent.
Artists in the US have also filed class action suits against Stability AI and other companies after discovering their work had been used to train AI systems.
The European parliament’s proposals, which are expected to be ratified by next month, could also include a ban on predictive policing, the use of algorithms to predict crime or identify suspects, as well as emotional recognition, using AI to detect human feelings and state of mind.
Those involved in the negotiations warned that it was too soon to say exactly what would make the text until the end of the discussions next week.
Dragoș Tudorache, an MEP and co-leading negotiator, said regulating AI was crucial in light of the growing popularity of generative AI products such as ChatGPT. “It is a pioneering technology and we need to harness it, which means putting rules in place,” he said. “Self due diligence by companies is not enough.”