EP Committees discuss the AI Act 

European Parliament’s Committees dig deeper into the difficult task of enforcing the EU’s landmark Artificial Intelligence Act (AI Act).  Last week, two parliamentary committees — LIBE and IMCO — convened to take stock of progress on the AI Act’s implementation. Lawmakers and Commission officials aired diverging views on whether the evolving Code of Practice is living up to the Act’s ambitious standards or quietly watering them down under pressure from industry.

At the LIBE Committee meeting on April 23, Irish MEP Michael McNamara (Renew), co-chair of the AI Working Group, updated colleagues on the group’s work since November. He cited six meetings held, three more planned before summer, and a year-end joint hearing with national parliaments. The group’s main focus: translating the AI Act’s prohibitions under Article 5 into clear guidance and finalizing a GPAI Code across four thematic subgroups.

Stakeholders such as BEUC, EDRi, and the EDPS have already weighed in, raising flags over legal clarity, copyright, and transparency. But tensions surfaced around how far the Code should go. McNamara acknowledged that open questions remain — particularly around fundamental rights — but emphasized close cooperation with the Commission’s AI Office and a commitment to timely progress.

However, some MEPs were not reassured.

German MEP Birgit Sippel (S&D) called for urgent implementation that prioritizes democratic principles. Jaroslav Bžoch (CZ, PfE) issued a stark warning: if the Code overreaches, it could become a regulatory straitjacket that stifles innovation and drives AI investment out of Europe.

“We risk creating more problems than we solve,” Bžoch said, arguing that the Code must remain a support tool, not a shadow regulation.

Belgian MEP Saskia Bricmont (Greens) took the opposite tack, defending the AI Act as a historic safeguard of rights and privacy. She urged greater EU investment in homegrown AI aligned with European values — not watered down by lobbying.

McNamara agreed that rights protections remain central, but reminded lawmakers that the pace is dictated by the expert drafters: “They are moving fast — but not faster than the law allows.”

Just five days later, the IMCO Committee picked up the thread — and turned up the heat.

Italian MEP Brando Benifei (S&D), the other Working Group co-chair, expressed dismay that the latest draft of the Code reflects an “industry-first” approach. Key obligations around societal risk assessments — including on disinformation and discrimination — are now framed as optional, he warned.

“This undermines the very spirit of the AI Act,” Benifei said, calling on the Commission to halt the drift and reassert the Act’s original balance.

Lucila Sioli, head of the Commission’s AI Office, responded with a defense of the process. The Code, she explained, remains voluntary and is now in its fourth draft, with public consultations ongoing. She promised clearer guidance, expanded definitions of systemic risk, and templates for training data transparency.

Yet skepticism remained.

MEPs Svenja Hahn (Renew) and Jaroslav Bžoch (PfE) both worried about a “second layer” of AI regulation emerging through the Code. Hahn warned of legislative overreach and procedural shortcuts, while Bžoch slammed the lack of business representation in the drafting process.

Greens MEPs Alexandra Geese and Kim Van Sparrentak countered that voluntary risk mitigation isn’t enough in the age of disinformation warfare. The Code, they insisted, must match the geopolitical moment and protect democratic systems.

Meanwhile, Renew MEP Sandro Gozi pressed the Commission on ensuring diversity — both in advisory boards and in the data used to train AI models. And EPP’s Andreas Schwab asked whether industry concerns stem from genuine regulatory excess or simply from delays in finalizing rules.

In closing, Sioli assured the committee that the Commission is working to “balance ambition with pragmatism.” Existing AI systems will have until 2027 to comply with the new rules, and work continues on aligning guidance with the AI Act’s letter — and spirit.

As the Code inches toward its final form, the message from Parliament is clear: Europe’s AI revolution is not just about algorithms — it’s about values, vigilance, and a vision for digital sovereignty.