EU AI Act Article 87 — protection of persons reporting infringements

What it does

Reports of AI Act infringements by workers in the broad sense covered by the Whistleblower Directive.

Honest assessment

This is the most substantive statutory AI whistleblower protection in force anywhere as of September 2026 — notably stronger than the US position, where the federal AI Whistleblower Protection Act remains stuck in committee. Its weakness is that it covers AI Act infringements, not 'AI is dangerous' concerns that break no specific provision.

How to file

Its single operative sentence: 'Directive (EU) 2019/1937 shall apply to the reporting of infringements of this Regulation and the protection of persons reporting such infringements.' In practice this means internal channels first, then the competent national authority, with external/public disclosure protected in defined circumstances. The AI Office's whistleblower tool is the EU-level external channel.

Format

As per national implementations of the Directive.

Timing

Applies from 2 August 2026.

What happens after

Confidentiality safeguards and anti-retaliation protection apply; remedies are pursued under national implementing law.

What it accepts

Reports of AI Act infringements by workers in the broad sense covered by the Whistleblower Directive.

What it does not accept

Does not create a new standalone AI whistleblower regime; it imports the existing Directive wholesale.

Operated by

European Commission / Member States, applying Directive (EU) 2019/1937