Make it easier for EU companies to develop and use AI

The EPP Group will vote today to delay and simplify the EU’s new rules on Artificial Intelligence (AI), so that companies can better prepare and face fewer overlapping requirements.

"We want to give companies more time, simplify the AI Act and support start-ups and scale-ups. Bureaucracy and duplicative rules are not good for people or businesses," said Arba Kokalari MEP, the Parliament’s negotiator on the file, ahead of today’s vote in the joint committees on the Internal Market and on Civil Liberties.

The proposal also aims to avoid situations where companies must comply with several sets of rules for the same product. Instead, existing industry rules should apply, so businesses only have to deal with one clear framework.

"Our goal is simple: fewer overlapping rules and lower costs for companies. This is what simplification should look like," said Axel Voss MEP, the EPP Group’s negotiator in the Civil Liberties committee.

Under the proposed timeline, key obligations for high-risk AI systems would apply later than originally planned, giving businesses more time to comply.

"If Europe wants to become an AI continent, we must make it easier, not harder, for companies to innovate and invest. Today, the rules may be too strict or not clear enough. This may discourage companies from adding useful AI features, even when they are safe. We should not treat low-risk AI like high-risk AI, or companies will stop innovating," Kokalari added.

The EPP Group also supports banning AI tools that generate fake nude images of real people without their consent. Recent cases of AI tools used online to 'undress' individuals in images highlight the urgent need for action. However, such a ban should focus only on abusive uses. It should not block legitimate uses of AI, such as photo editing, healthcare applications, or tools used in retail, arts and fashion.

Negotiations with EU countries are expected to conclude by April, with final adoption before August 2026, to ensure that strict rules for high-risk AI do not end up applying to low-risk applications.