The EU Just Moved Its AI Deadlines. Here Is the New Calendar
Europe delayed its high-risk AI rules by more than a year, tightened one transparency deadline and added new bans. What changes if you sell into the EU.
Rules for high-risk AI systems under the EU AI Act were due to apply on August 2, 2026. They did not. In June 2026 the Council gave final approval to the Digital Omnibus on AI, a package of targeted amendments that pushes those obligations back and links them to the availability of the standards companies need to comply.
Key EU AI Act dates after the Digital Omnibus on AI.
What moved
- Stand-alone high-risk AI: December 2, 2027: Covers use cases such as hiring and worker management, creditworthiness, some insurance underwriting, education and access to essential services.
- High-risk AI inside products: August 2, 2028: AI embedded in regulated products such as medical devices, toys or machinery gets an extra year.
- National AI sandboxes: December 2, 2027: The deadline for member states to set up regulatory sandboxes was also postponed.
What got stricter
Not everything slowed down. The grace period for providers to label AI-generated content was cut from six months to three, setting a new deadline of December 2, 2026. The law also adds a new prohibited practice: using AI to generate non-consensual intimate content or child sexual abuse material.
- 16 mo: Delay for stand-alone high-risk rules
- Dec 2026: AI-content labelling deadline
- 234: Signatories to the GPAI Code of Practice
On general-purpose AI, the voluntary Code of Practice now has 234 signatories, including major model providers such as Anthropic, Google, Meta, Microsoft, Mistral and OpenAI.
For Canadian companies If you sell AI-powered products or services into the EU, the delay buys time, not an exemption. Use it to map which of your systems would count as high-risk and start the documentation now.
A practical checklist
- List every AI system you offer to EU customers and note what decisions it influences.
- Flag anything touching hiring, credit, insurance, education or essential services as potentially high-risk.
- Check how your products label AI-generated content before the December 2026 deadline.
- Ask your AI vendors whether they have signed the GPAI Code of Practice.