EU AI Act Timeline 2024–2027: Every Key Date

The EU AI Act entered into force on 1 August 2024 and applies in phases: prohibited practices and AI literacy from February 2025, general-purpose AI obligations from August 2025, transparency and most remaining rules from August 2026, and certain high-risk product obligations from August 2027.

August 2024 — entry into force

The Regulation entered into force, starting the clock on every later deadline. Nothing is directly enforceable yet, but the definitions, roles and risk tiers become the vocabulary every later obligation is written in. Organisations that mapped their provider/deployer roles early had far less to redo afterwards.

February 2025 — prohibitions and AI literacy

The bans on unacceptable-risk practices became applicable, alongside the duty for providers and deployers to ensure a sufficient level of AI literacy among staff operating AI systems. This is already live. It is also the phase most often missed, because it has no product deliverable attached.

August 2025 — general-purpose AI

Obligations for general-purpose AI models started applying, together with governance structures and the penalty regime for those duties. Codes of practice provide the operational detail while harmonised standards are still being finalised. Downstream deployers feel this indirectly, through the documentation and marking they receive from upstream providers.

August 2026 — transparency and the general regime

The bulk of the Regulation, including the Article 50 transparency obligations for synthetic content and deepfakes, becomes applicable. This is the date that matters for anyone publishing AI-generated or AI-modified media. Marking and disclosure need to be running in production before this point, not designed at it — and because standards land late, the mechanism should be replaceable without re-engineering the whole pipeline.

August 2027 — high-risk in regulated products

The extended deadline for high-risk AI embedded in products already covered by EU product-safety legislation. Conformity assessment, technical documentation and post-market monitoring dominate this phase. Evidence built for transparency earlier is reusable here: the same signed, time-anchored records feed technical documentation and audit trails.

How to sequence AI Act readiness

Confirm roles per system

Record whether you are a provider, deployer, importer or distributor for each AI system in use.

Close the literacy gap

Train the teams that operate AI systems — this obligation is already applicable.

Collect upstream documentation

Ask model providers for the transparency and marking information you will need to pass downstream.

Ship marking before the deadline

Put machine-readable marking and disclosure into the publishing pipeline ahead of the transparency date.

Keep durable evidence

Retain independently checkable records of each declaration so later phases can reuse them.

Frequently asked questions

Is the EU AI Act already in force?

Yes. It entered into force in August 2024 and its obligations apply in stages; prohibitions, AI literacy and general-purpose AI duties are already applicable.

When do the transparency rules for AI-generated content apply?

The Article 50 transparency obligations apply from August 2026, together with the bulk of the Regulation.

Do the deadlines change if standards are late?

Application dates are set by the Regulation. Missing harmonised standards affect how compliance is demonstrated, not whether the obligation applies.

Does the timeline apply to non-EU companies?

Yes, where output is placed on the EU market or reaches people in the EU, regardless of where the organisation is established.

Get transparency evidence in place early

See what a machine-readable declaration looks like before the deadline.