Radar
Upcoming compliance deadlines across the AI governance landscape.
- 2 Dec 2026EU
Deadline for AI-content watermarking solutions
Under the Digital Omnibus agreement, the grace period for implementing technical transparency solutions for artificially generated content ends, with compliance due by this date.
Content and product teams shipping generative features in the EU need machine-readable marking implemented, not just planned.
- 1 Jan 2027Colorado, US
Revised Colorado AI law and Chatbot Safety Act take effect
SB 26-189 replaces Colorado's original AI Act with a narrower framework: notice requirements, an adverse-action and human-review process, and three-year record retention for automated decision-making in consequential decisions. The Chatbot Safety Act takes effect the same day.
Employers and lenders using automated decision tools for Colorado residents need notice and human-review workflows ready; chatbot operators face age-estimation and disclosure duties.
- 22 Jan 2027South Korea
AI Basic Act grace period ends: enforcement begins
South Korea's AI Basic Act (formally the Framework Act on the Development of Artificial Intelligence and the Establishment of a Foundation for Trustworthiness) took effect on 22 January 2026, but MSIT is running a grace period of at least one year during which administrative fines are deferred, except in cases involving serious social harm. Substantive obligations already apply; the enforcement teeth arrive at the end of the grace window. The Act applies extraterritorially.
Organisations serving Korean users need transparency notices, generative-AI labelling, and, where thresholds are met, a designated local representative in place before fines become live.
Source → - 2 Dec 2027EU
AI Act high-risk obligations apply (Annex III)
The deferred core of the AI Act: full provider and deployer obligations for stand-alone high-risk systems in areas like employment, credit, education, and essential services.
Multinationals with EU operations get sixteen extra months: an extension of time, not a relaxation of the underlying obligations.
- 2 Aug 2028EU
High-risk obligations for AI embedded in regulated products
Obligations extend to AI systems that are products or safety components under EU product-safety law (medical devices, machinery, and similar), with a narrowed definition of 'safety component'.
Manufacturers embedding AI in regulated hardware inherit conformity-assessment duties on this later track.