AI Regulation Is Here. Almost. EU lawmakers move forward with draft law to regulate development of AI-powering products such as ChatGPT

BRUSSELS—Advances in artificial intelligence this year have rocked the tech industry, triggering calls from politicians, consumer groups and AI executives themselves for rules governing how to use the technology.

Those regulations are now taking shape , at least on this side of the Atlantic. The European Union's parliament voted Wednesday to push forward draft legislation, called the AI Act, that is positioned to be the West's first comprehensive set of AI regulations.

The draft rules include bans on real-time, remote biometric surveillance in public spaces and would prohibit harvesting surveillance footage or scraping the internet in developing facial-recognition databases. The parliament's version also seeks a ban on so-called predictive policing systems, which analyze prior criminal behavior and other data and try to predict future illegal activity.

More broadly, the draft legislation aims to regulate how companies train AI models with large data sets. It would, in some cases, require companies to disclose when content is generated using AI.

Under the rules, companies would also need to design their AI models in a way that prevents them from creating illegal content, and they would be required to publish summaries of the copyrighted data used to train their models.

Such an obligation would give publishers and content creators a potential means to seek a share of profits when their works are used as source material for AI-generated content by tools such as ChatGPT.

Current drafts of the bill would impose fines of up to 6% or 7% of a company's global revenue in certain cases of noncompliance.

The parliament's version of the legislation passed Wednesday will now form the body's negotiating position when talks to determine the final shape of the legislation begin among representatives from the parliament, EU member states and the European Commission. Officials have said they plan to launch those negotiations immediately and aim to reach a deal on the proposed law before the end of this year.

Tech companies and their lobbyists argue that any government-enforced rules should focus on specific AI applications—and not put too many restrictions on how AI is developed, as is being proposed in Europe. They say such an approach would impede innovation.

But some tech researchers have joined academics and technologists in expressing support for rules like those being formulated in the EU, that could effectively slow down a race by companies to roll out advanced new AI tools by regulating how such tools are developed in the first place.

Earlier this year, a group of AI researchers and tech executives including Elon Musk signed an open letter that called for a six-month moratorium on the training of the next generation of AI tools to give time for regulators and industry to set safety standards. Last month, a group of researchers said mitigating risks of human extinction from AI should be a global priority.

The AI legislation was proposed in 2021 by the European Commission, the bloc's executive body. The push to set out rules for AI has taken on new urgency over recent months after the release of tools such as ChatGPT. The tool, developed by Microsoft-backed startup OpenAI, can respond to users' written questions.

European officials hope the proposed legislation will be a world-first that can set the standard for other jurisdictions and for the companies that make and use the technology.

The rapid development of AI in recent months has prompted governments worldwide to consider whether to adopt new rules for powerful AI tools. China's top internet regulator proposed draft rules in April and the Biden administration is looking at whether checks are needed.

The EU's legislation "will set the tone worldwide in the development and governance of artificial intelligence," EU lawmaker Dragos Tudorache said. The Romanian lawmaker led the institution's work on the AI legislation with Italian lawmaker Brando Benifei.

The Computer & Communications Industry Association, a lobby group, said some of the provisions backed by the parliament risk creating overly prescriptive rules for relatively low-risk AI applications and hindering innovation. Consumer groups have said the bans proposed by the parliament are needed to protect people's fundamental rights.

EU officials have sought to position themselves as front-runners in setting up guardrails on AI systems that they say should support innovation while limiting the biggest risks of the technology.

The EU legislation "is about acting fast and taking responsibility," said Thierry Breton, the EU's internal market commissioner.

Write to Kim Mackrael at [email protected]

Credit: By Kim Mackrael