Regulation (EU) 2024/1689  ·  For UK businesses
← Back to blog
EU AI Act 25 June 2026

High-risk AI Act deadlines delayed — what UK businesses need to know

The European Parliament approved significant amendments to the EU AI Act on 16 June 2026. The main change: high-risk AI obligations have been pushed back by up to 16 months. For UK businesses with EU customers or employees, this is welcome breathing room — but it does not mean you can stop preparing.

What happened

On 16 June 2026, the European Parliament formally approved a package of amendments to Regulation (EU) 2024/1689 — the EU AI Act. The amendments are part of a broader "Digital Omnibus" package aimed at simplifying EU digital regulation and reducing compliance burdens, particularly for smaller businesses.

The Council of the EU is expected to formally adopt the amendments by late July 2026, with publication in the Official Journal expected to follow shortly after. The changes are intended to be in force before 2 August 2026, when several key provisions would otherwise have applied.

The key change: high-risk deadlines pushed back

The most significant change for UK businesses is the extension of compliance deadlines for high-risk AI systems listed in Annex III of the Act — the category that covers AI used in employment, recruitment, credit scoring, education, and essential services.

2 August 2026
Transparency obligations (Article 50) apply from this date. If your business uses AI that interacts directly with people — chatbots, virtual agents, AI-generated content — you must ensure users are clearly informed they are interacting with AI.
2 December 2027
High-risk AI obligations (Annex III) now apply from this date — delayed by 16 months from the original 2 August 2026 deadline. This covers AI used in recruitment, employee monitoring, performance assessment, credit scoring, and educational evaluation.
2 December 2026
New prohibition on nudifier applications. AI tools that generate non-consensual intimate imagery or child sexual abuse material are now explicitly prohibited under Article 5.
2 August 2028
High-risk AI embedded in regulated products (medical devices, machinery, automotive) face obligations from this date — a 12-month delay from the original August 2027 deadline.

What stays the same

Not everything has been delayed. Several obligations are unchanged:

  • Prohibited AI practices (Article 5) have been in force since 2 February 2025. Social scoring, subliminal manipulation, real-time biometric identification in public spaces, and emotion inference in workplaces and schools are already banned.
  • Transparency obligations (Article 52/50) apply from 2 August 2026 as planned. If your business uses customer-facing AI chatbots or generates AI content, disclosure requirements are unchanged.
  • General-purpose AI model obligations have been in force since August 2025.

What this means for UK SMEs specifically

If your business uses AI tools such as Workday, SAP SuccessFactors, or LinkedIn Recruiter to screen job candidates, assess employee performance, or make decisions that affect EU-based individuals, those tools fall into the high-risk Annex III category. The 16-month delay gives your business more time to prepare — but the obligations have not gone away.

The delay reflects a practical reality: the harmonised standards and supporting guidance that businesses need to actually comply were not going to be ready in time. Rather than enforce against unprepared businesses, the EU has chosen to align deadlines with when the necessary tools and standards will exist.

What your business should do now
  • Use the extended time to complete a full AI software inventory — our compliance tool can help with this.
  • Identify which of your AI tools fall into the high-risk Annex III categories, particularly those used in recruitment, HR, and performance management.
  • Check that any customer-facing AI (chatbots, virtual agents) is already disclosing its AI nature — this obligation applies from 2 August 2026.
  • Do not wait until late 2027. The documentation, human oversight procedures, and risk assessments required for high-risk AI take time to build.

Sources

This post summarises publicly available regulatory developments for general information only. Nothing here constitutes legal advice. For advice specific to your business, please consult a qualified legal professional.