Europe's AI Act is no longer distant policy. By our calendar, there are just 53 days until the August 2, 2026 milestone when broad transparency obligations for interactive AI systems begin to bite.
The timeline has not stood still. The late-May 2026 Omnibus amendments reshuffled several deadlines, added grace periods for high-risk and general-purpose models, and introduced additional prohibitions. For B2B SaaS teams in the EU market, this is a call to execute, not to pause.
A Sleeping Market Wakes Up
A year ago, many compliance teams treated the AI Act as a future problem. That has changed quickly. We are seeing materially higher scan activity from teams asking the same operational questions: what is our risk tier, what gaps remain, and can we publicly show remediation progress.
This urgency is rational. Unacceptable-risk prohibitions are already in force, baseline GPAI obligations started earlier, and August 2, 2026 marks the next market-wide transparency checkpoint for interactive AI.
What Changed in the May 2026 Omnibus Package
The Council's corrections package was technical in tone but highly practical in impact. Three updates matter most for product teams:
- Article 50(2) synthetic-content labelling for GPAI providers shifted from August 2, 2026 to December 2, 2026.
- Annex III high-risk system compliance moved to December 2, 2027, and Annex I embedded product timelines moved to August 2, 2028.
- Generation of non-consensual intimate imagery was explicitly added to prohibited practices.
Extra time exists for some obligations, but not for governance maturity. Classification, documentation, and operational controls still need to be built now.
Risk Categories in Practical Terms
Unacceptable Risk
Prohibited practices under Article 5, including specified manipulative and harmful use classes.
High Risk
Systems in sensitive domains with strict controls, conformity assessment, registration, and oversight requirements.
Limited Risk
Interactive services with transparency duties, including clear user disclosure of AI interaction.
Minimal Risk
Low-impact use with limited mandatory obligations, plus voluntary codes and best practices.
Providers of general-purpose models still carry cross-cutting obligations such as technical documentation, training-data transparency posture, and risk controls that sit outside a simple four-tier framing.
Why This Matters for B2B SaaS
Many SaaS products now embed AI by default, sometimes without teams explicitly naming those modules as regulated AI systems. In procurement, that ambiguity creates friction.
- Classify early. If AI materially shapes outcomes in hiring, credit, infrastructure, or similar domains, prepare on a high-risk assumption.
- Document continuously. Transparency notices are only the start; technical files and evidence trails must stay current.
- Operationalize oversight. Human intervention, incident processes, and logging are expected, not optional.
- Harden abuse controls. Generative workflows need immediate safeguards against prohibited harmful outputs, including NCII.
How ComplianceRadar Helps
We built ComplianceRadar to support this exact reality: moving deadlines, evolving interpretations, and practical remediation workflows.
Teams can track finding status, monitor progress, publish verification views, and maintain consistent trust-language while preserving legal-safe phrasing for customer-facing transparency.
Instead of treating compliance as a static PDF event, the platform works as a living system for gap tracking, evidence management, and ongoing readiness communication.
Conclusion: Don't Waste the Extra Time
The next hard checkpoint remains close. Limited-risk transparency obligations still take effect on August 2, 2026. GPAI synthetic-content-labelling obligations now follow in December 2026, with high-risk milestones in 2027 and 2028.
Strong teams are using this window to classify systems, build risk registers, prepare documentation, and engage auditors before pressure peaks.
If you are unsure where to begin, start by cataloging AI features across your portfolio and mapping them to Annex III exposure. Then track remediation visibly and continuously.
Countdowns are useful only if they trigger action. Every day you delay evidence and governance work, your August and December timelines get tighter.
Start your 53-day readiness sprint
Scan your AI product surface, identify regulatory gaps, and track remediation progress with a living compliance workflow.
Sources and further reading
- Regulation (EU) 2024/1689 (EU AI Act) — EUR-Lex
- EU AI Act Updates: Delays for High-Risk Systems and the Ban on Nudifier Apps
- 99 Days Until the EU AI Act Deadline
- The EU AI Act Audit Process
This article is informational and does not constitute legal advice.