What is the new compliance deadline for high-risk AI in the EU?

Under the AI Omnibus agreement of May 2026, the compliance deadline for most high-risk AI systems affecting fundamental rights is 2 December 2027. AI embedded in regulated products has until 2 August 2028.

Which AI applications are considered high-risk under the EU AI Act?

High-risk AI includes systems used in employment (CV screening), essential services (credit scoring), education (admissions), biometrics, and critical infrastructure management.

Do SMEs have different AI Act obligations?

Yes. The AI Omnibus provides lighter obligations and reduced fees for SMEs and small mid-caps. AI literacy requirements have also been adjusted to a 'best efforts' standard for these entities.

When do AI transparency and watermarking rules take effect?

Transparency and watermarking requirements apply from August 2026, though AI systems already on the market have until 2 December 2026 to comply.

EU AI Omnibus: New Compliance Deadlines for Maltese Businesses

May 8
2 min read

Key Takeaways

Extended Deadlines: High-risk AI systems now have a compliance window until December 2027 or August 2028, depending on the category.
Immediate Obligations: Transparency and watermarking requirements remain set for August 2026.
High-Risk Focus: Businesses using AI for HR, credit scoring, or essential services must meet strict EU AI Act standards.
SME Support: Small and medium-sized enterprises receive lighter obligations and "best efforts" standards for AI literacy.

infographic on the AI Omnibus requirements

Political Agreement Resets AI Act Timeline - AI Omnibus

On 7 May 2026, EU legislators reached a political agreement on the AI Omnibus, a regulation amending the original EU AI Act (Regulation (EU) 2024/1689).

This amendment addresses concerns that the original August 2026 deadlines were unworkable

due to pending technical standards.

Revised AI Compliance Schedule

The AI Omnibus establishes a tiered timeline for different AI applications:

Fundamental Rights (High-Risk): AI systems used in areas like HR, education, and law enforcement must comply by 2 December 2027.
Regulated Products: AI embedded in medical devices, machinery, and toys has a deadline of 2 August 2028.
Transparency & Watermarking: Requirements for AI-generated content still apply from August 2026, with an extension to December 2026 for systems already on the market.

Defining High-Risk AI Systems

Under the EU framework, AI is classified as "high-risk" if it significantly impacts safety or fundamental rights. Key categories include:

Employment & HR: Systems for screening CVs, ranking candidates, or monitoring performance.
Access to Essential Services: AI used for credit scoring, insurance pricing, or healthcare triage.
Education: Tools assessing admissions or student learning outcomes.
Critical Infrastructure: Management of water, energy, or traffic.
Biometrics: Remote facial recognition or emotion detection.

For a full list of high-risk categories, refer to Annex III of the AI Act.

Impact on Maltese SMEs and Mid-Caps

The AI Omnibus extends specific privileges to both SMEs and small mid-cap companies.

Lighter Obligations: Reduced fees and simplified compliance paths for smaller entities.
Proportionate AI Literacy: The mandate to ensure staff understand AI tools has shifted from a strict legal requirement to a "best efforts" standard.

New Prohibitions: AI "Nudifiers"

Effective December 2026, the EU will ban AI tools designed to generate non-consensual intimate imagery. This signals a shift toward regulating AI at the model level to prevent societal harm.

Strategic Recommendations for Organizations

The AI Omnibus provides a reprieve, not an exemption. Maltese businesses should utilize this window to:

Conduct an AI Audit: Inventory all AI tools currently in use.
Assess Risk Levels: Determine which tools fall under the "high-risk" classification.
Establish Governance: Build internal frameworks for AI oversight and literacy.

References:

Tech Policy Press (8 May 2026)
European Commission Press Corners IP/26/1024 & IP/25/2718
European Parliament Press Room

Transparency Disclosure: AI-Assisted Content

This article, including any images, was generated with the assistance of a Large Language Model (LLM) but has undergone a comprehensive process of human review and editorial control. In accordance with the exceptions outlined in Article 50(4) of the EU AI Act and the draft Code of Practice, this publication is subject to the editorial responsibility of Synerf. The review process involved verifying factual accuracy, ensuring contextual relevance, and exercising organizational oversight to maintain the integrity of the information provided.