Crunch time for the AI omnibus in the European Parliament. Progress on AI literacy, sensitive data and enforcement, but legal timelines, a sexual deepfake ban and regulatory privileges are “parked” for political talks. My latest for MLex.
www.mlex.com/mlex/article...
India's AI Impact Summit exposed quiet EU unease over proposed “deliverables,” including a new AI inclusion alliance and AI-for-science network, seen as duplicating UN structures. A compromise avoided a diplomatic rift just after the EU-India trade deal.
www.mlex.com/mlex/article...
EU countries will discuss tomorrow adding two new bans to the AI Act via the omnibus. Draft proposals would prohibit AI systems used to generate non-consensual sexual deepfakes and child sexual abuse material. Support is growing, but questions remain.
www.mlex.com/mlex/article...
The second Council compromise on the AI omnibus refines the new AI Office's enforcement powers, incl. due process safeguards, introduces an EDPS-led sandbox, reinforces existing AI literacy duties, and sets a deadline for post-market guidance.
www.mlex.com/mlex/article...
MEPs broadly agree that sexual deepfakes should be tackled in the AI Act omnibus, but they’re split on how. Renew + S&D back a ban limited to non-consensual content. Greens want an outright ban. S&D also proposes duties on GenAI providers.
www.mlex.com/mlex/article...
The Commission has set its 2026 priorities for AI Act's implementing and delegated acts, incl. GPAI enforcement rules & the AI Office’s market surveillance powers. Energy consumption methodology and systemic risk criteria are deprioritized.
www.mlex.com/mlex/article...
More on MLex.
Leading MEPs are backing fixed AI Act deadlines, dropping the Commission’s discretion to anticipate high-risk duties. The draft report restores AI literacy duties & tightens sensitive data use, while leaving tougher issues like registration for later.
www.mlex.com/mlex/artific...
Momentum is building to add a ban on sexualized deep fakes to the EU’s AI Act. After the Grok scandal, EU lawmakers were the first to call for a ban. Now Spain, France, Ireland, and Slovenia are backing a similar move in the Council. But legal limits loom.
www.mlex.com/mlex/article...
Centralizing AI Act enforcement is emerging as a key fault line in the AI omnibus talks. EU governments are pushing back against expanding the AI Office’s powers, setting up a likely clash with the European Parliament.
www.mlex.com/mlex/article...
EU countries are pushing to keep the AI Act’s high-risk regime narrow for law enforcement. As the Commission drafts guidance, 23 of 27 member states back a letter warning that a broad interpretation would undermine police capabilities & slow investigations
www.mlex.com/mlex/article...
The Council’s first compromise on the AI omnibus is out, and it pushes back hard on the Commission’s original approach. The text favors fixed deadlines for high-risk AI obligations & reinstates a simplified registration requirement, among other changes.
www.mlex.com/mlex/artific...
France is quietly testing support among EU capitals to split the AI “omnibus,” separating the delay of high-risk duties from broader changes. The aim is to reopen space for deeper reform, but some countries fear it could bog talks down and derail adoption.
www.mlex.com/mlex/article...
EU Commission plans to expand cybersecurity certificates to cover companies’ overall risk-management posture, a draft of the revamped Cybersecurity Act shows. The reform aims to revive stalled EU cyber certification by introducing clearer procedural rules.
www.mlex.com/mlex/article...
The Commission will specify in upcoming guidelines what documentation AI developers must prepare when relying on the AI Act’s so-called “filter” to conclude that their system does not qualify as high-risk.
www.mlex.com/mlex/article...
The Frontier AI initiative is taking shape. Backed by 🇫🇷, 🇩🇪 & the Commission, it aims to create a nonprofit lab in Europe to support frontier AI research not yet viable for markets. An internal paper points to early support by 30 leading AI researchers.
www.mlex.com/mlex/article...
The Cyprus presidency aims to secure a negotiating mandate on the AI omnibus by early April, after initial talks pointing to limited changes. The main exception is the plan to drop the registration requirement, which is facing pushback from 🇪🇸 & 🇩🇪
www.mlex.com/mlex/article...
The draft revamp of the Cybersecurity Act that went to interservice consultation includes a mandatory phase-out of Chinese ICT equipment from critical sectors. The proposal faces strong resistance inside the Commission and from Spain & Germany
www.mlex.com/mlex/article...
Some end-of-the-year news from the Commission's AI Office. After one and a half years, Matthieu Delescluse is set to be appointed head of unit for AI Safety, a key role for enforcing the AI Act's rules on GPAI models.
www.mlex.com/mlex/article...
The European Parliament's report on algorithmic management sends mixed messages on whether MEPs want binding rules, and is unlikely to break the deadlock inside the Commission on whether this topic should be addressed in the upcoming Quality Jobs Act.
www.mlex.com/mlex/article...
The EU Parliament has ended its internal dispute over who will steer the AI Act changes, confirming joint IMCO-LIBE lead. Focus now shifts to who will be rapporteur, with Brando Benifei challenging the EPP’s bid to give the lead to Arba Kokalari in IMCO.
www.mlex.com/mlex/article...
The Commission’s health package proposes moving MDR & IVDR from Annex I Section A to B under the AI Act. A technical tweak with big consequences, shifting from immediate high-risk compliance to a sector-driven approach via secondary acts.
www.mlex.com/mlex/article...
The Commission is weighing a plan B for the AI Act delay. If the omnibus stalls ahead of the August 2026 deadline, a fallback option being considered is to carve out the postponement of high-risk rules into a separate fast-track proposal, MLex has learned.
www.mlex.com/mlex/article...
France wants common specs for the development of labelling and detection tools for AI-generated content, as a fragmented landscape of proprietary solutions would leave Europe dependent on foreign tech in areas such as justice, journalism, and elections.
www.mlex.com/mlex/article...
The Commission gambled by tying the AI Act’s high-risk delay to the full AI omnibus. For now, it isn’t paying off. Parliament is stuck in a new turf war as ITRE & JURI challenged the IMCO-LIBE lead, & some rapporteurs won't be appointed until late January.
www.mlex.com/mlex/article...
The Commission’s plan to delay the AI Act’s high-risk regime is facing serious questioning from EU governments. Countries are questioning the Commission’s discretion to set new deadlines and demanding clear criteria for predictability.
www.mlex.com/mlex/article...
BREAKING: It's game over for the AI Liability Directive. In a decisive vote, the European Parliament’s Legal Affairs Committee rejected (by 22 votes to 1) the proposal to legally challenge the European Commission’s withdrawal of the file.
www.mlex.com/mlex/article...
Will the Commission propose binding rules on AI in the workplace? Today’s Quality Jobs Roadmap won’t say. A draft seen by MLex keeps the door open but narrows the path, amid internal political tensions.
www.mlex.com/mlex/article...
The Commission is preparing an implementing act detailing how the AI Act will be enforced for GPAI models. It will cover model evaluations, investigative procedures, rights of providers, and the sanction regime, drawing on DMA/DSA enforcement.
www.mlex.com/mlex/article...