The EU has agreed to simplify the EU AI Act. After many hours of negotiations last week, they signed off on an extended deadline for high-risk AI systems, removed machinery from the Act, and added a ban on AI-generated nude images.
The agreement last week is part of the European Commission’s push to simplify AI rules to promote innovation. The simplification drive came after businesses complained about overlapping regulations and red tape that they said hampered their ability to compete with US and Asian rivals.
The EU AI Act is the first comprehensive AI law in the world. It entered into force in August 2024, with key provisions phased in over time. The agreement reached on Thursday still needs formal approval from EU governments and the European Parliament in the coming months, but the political deal is done.
Marilena Raouna, Cyprus’s deputy minister for European affairs, said in a statement, “Today’s agreement on the AI Act significantly supports our companies by reducing recurring administrative costs.” Cyprus currently holds the rotating EU Council presidency.
The European Consumer Organisation says the changes weaken the Act. The tech lobbying group CCIA says lawmakers should have gone further. So the criticism comes from both sides.
The high-risk delay
Here is the headline change. The rules governing high-risk AI systems, those involving biometrics, critical infrastructure, and law enforcement, will now apply from the 2nd of December 2027. The previous deadline was the 2nd of August this year. That is a 16-month extension.
What does high-risk actually mean? Under the AI Act, high-risk systems include AI used in recruitment, education, critical infrastructure, law enforcement, migration and border control, the administration of justice, and biometric identification. If you are deploying AI in any of these areas, the strict obligations around risk assessment, data quality, human oversight, and post-market monitoring now have a new deadline.
For organisations, this is more breathing room, not a free pass. The technical standards needed to comply are not yet finalised. The European Commission says these will be published later this year. So the delay reflects the reality that nobody had the tools ready to comply on time.
Machinery carved out
The second change is more technical but worth flagging. Machinery has been excluded from the AI Act entirely, on the basis that it is already covered by separate sectoral rules. This came after pressure from major industrial players, including Germany’s Siemens and Dutch chipmaker ASML, the same ASML that is the largest shareholder in Mistral AI. If your organisation makes or deploys industrial machinery with AI components, this is good news. The compliance burden is now governed by existing machinery directives, not by the AI Act on top.
The deepfake ban
The deal includes a ban on AI practices that create unauthorised sexually explicit images. This applies from the 2nd of December this year. It is a direct response to AI-generated nudification, including content produced by Elon Musk’s xAI chatbot Grok on X, which I covered in detail in episode 283 last year.
Dutch lawmaker Kim van Sparrentak said, “By the end of this year, everyone, but especially women and girls, will be safe from horrific nudifier apps being widely available on the EU market.”
This is the EU acting fast on a specific harm where public outrage was loud, and the evidence was overwhelming. It also sets a precedent. When regulators see clear consumer harm, the timelines move quickly. When the harms are diffuse or technical, the timelines slip.
Mandatory watermarking from December
Linked to the deepfake ban, mandatory watermarking of AI-generated output applies from the 2nd of December 2026. This is a transparency requirement. If your organisation produces synthetic content, AI-generated images, video, audio, or text published to inform the public, you will need to mark it. This aligns with what I covered in episode 279 on disclosing AI-generated images. It is now law, not best practice.
What this means for your AI strategy
Three takeaways for anyone planning AI deployment this year.
First, do not assume the regulatory landscape is settled. The Commission proposed simplification in November 2025, and we are watching it being negotiated in real time. Stay informed. The rules on the books today may not be the rules in twelve months.
Second, if you are deploying high-risk AI, you have an extra 16 months. Use that time to build governance structures, document your risk assessments, and prepare for compliance. Do not wait until December 2027 to start.
Third, the deepfake ban and watermarking requirements will apply from this December. That deadline is real, and it is close.
The AI Act is still considered the strictest AI law in the world, even after this simplification. That is the Reuters line, and I think it holds. Europe is balancing competitiveness against accountability, and that balance is being struck in public, with industry, civil society, and lawmakers all in the room.
For your organisation, the practical question is not whether to wait for clarity. It is how to build governance now, knowing the rules will keep moving. That is the work.
If you want to build your AI literacy and skills, I can help
📚 Grab AI Playbook on Amazon, the perfect podcast companion.
📰 Subscribe to Human in the Loop, my LinkedIn Newsletter
🎓 Take the CPD Accredited AI Playbook Diploma
📞 Book a call at digitaltraining.ie
Sources
- Reuters via Insurance Journal, EU Countries, Lawmakers Clinch Provisional Deal on Watered-Down AI Rules, Foo Yun Chee, 7 May 2026: https://www.insurancejournal.com/news/international/2026/05/07/868855.htm
- European Commission, Shaping Europe’s digital future, AI Act page: https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai
- European Commission, European approach to artificial intelligence: https://digital-strategy.ec.europa.eu/en/policies/european-approach-artificial-intelligence
- Amnesty International, How EU proposals to ‘simplify’ tech laws will roll back our rights, 2 April 2026: https://www.amnesty.org/en/latest/campaigns/2026/04/eu-simplification-laws/
- European Commission, Digital Omnibus on AI Regulation proposal, 19 November 2025: https://digital-strategy.ec.europa.eu/en/library/digital-omnibus-ai-regulation-proposal
- AI SIX Episode 283, Grok’s Nudity Scandal: https://aisix.buzzsprout.com/
- AI SIX Episode 279, When to Use AI Images: https://aisix.buzzsprout.com/
Key Statistics
- Provisional deal reached after 9 hours of negotiations, 7 May 2026 (Reuters)
- High-risk AI rules delayed from 2 August 2026 to 2 December 2027, a 16-month extension
- Machinery is excluded from the AI Act entirely, after pressure from Siemens and ASML
- Ban on AI-generated unauthorised sexually explicit images applies from 2 December 2026
- Mandatory watermarking of AI-generated output applies from 2 December 2026
- AI Act entered into force in August 2024, full application originally 2 August 2026
- Commission Digital Omnibus proposal published 19 November 2025
- AI Act remains the strictest AI law in the world, even after the changes (Reuters)
About the Podcast Host
AI and Communications Specialist | Author | Podcaster | International Speaker
Joanne Sweeney is a recognised leader in digital marketing and digital transformation, specialising in the practical application of AI to enhance communication strategies. The founder of Digital Training Institute, Joanne, empowers leaders and communications professionals to leverage AI responsibly and effectively, ensuring transparency, trust, and tangible results.
She completed the Harvard Kennedy School Executive Programme in AI and Public Policy this summer before launching her fourth book AI Playbook, in September. Joanne has also published two editions of Public Sector Marketing Pro, which were preceded by her first title, Social Media Under Investigation Law Enforcement and the Social Web.
Her books have become go-to resources for governments, public sector and policing agencies, as well as marketing and communications professionals, all navigating the challenges of the changing Internet.
Joanne is the host of the AI SIX Podcast, a daily six-minute podcast that provides actionable insights, expert interviews, and practical advice to help professionals navigate the new world of work.
