AI Regulatory Intelligence — by YRproject

factual analysis · traceable to primary sources

Analysis

GPAI enforcement goes live on 2 August 2026 — and the Signatory Taskforce

Adopted 2026-06-20 · ≈ 3 min read · Dirk Baaijen

From 2 August 2026 the European Commission can enforce the GPAI model rules, fines included (Art. 101). Obligations have applied since 2 August 2025; older models comply by 2 August 2027. The Signatory Taskforce, chaired by the AI Office, steers the Code of Practice.

Short answer: The obligations for providers of general-purpose AI (GPAI) models have applied since 2 August 2025, but the European Commission could not yet enforce them. That changes on 2 August 2026: from that date the Commission's enforcement powers — including the fines of Article 101 — enter into application. In the run-up, the AI Office has built the institution that will carry day-to-day oversight of the GPAI Code of Practice: the Signatory Taskforce, which held its constitutive meeting on 30 January 2026.

Three dates, one transition

The GPAI regime is phased in over three years, and the middle date is the one that matters in 2026:

DateWhat applies
2 August 2025Obligations for providers of GPAI models enter into application (Arts 53, 55).
2 August 2026The Commission's enforcement powers — requests for information, model evaluations, and fines — enter into application.
2 August 2027Providers of GPAI models placed on the market before 2 August 2025 must be in compliance.

So for a year the duties existed without a sanction behind them. From 2 August 2026 the AI Office can act: it can demand information, evaluate models, and, where warranted, impose fines of up to 3% of worldwide annual turnover or EUR 15 million, whichever is higher (Article 101). That is a separate track from the national fine regime of Article 99, which the fines and enforcement entry sets out. Providers whose models were already on the market before the obligations began get the longest runway — until 2 August 2027 — but no exemption.

The Signatory Taskforce: who watches the Code

The GPAI Code of Practice is the voluntary route (Article 56) by which providers demonstrate they meet their duties. A code is only as good as its application, and in 2026 the AI Office gave the signatories a standing forum for exactly that. On 30 January 2026 the signatories held the constitutive meeting of the Signatory Taskforce and adopted its rules of procedure by consensus. The Taskforce is chaired by the AI Office and meets at the frequency its work requires — at least once a year; by spring it had met three times (30 January, 13 March and 27 March 2026).

Its mandate is coordination rather than rule-making. According to the Commission, the Taskforce exists to facilitate exchange between signatories on implementing the Code, to provide input on guidance documents without prejudice to the public consultations the AI Office must run, to exchange views on technological developments, and to discuss relevant research and independent expert input. For public accountability the AI Office has published a Vademecum — the rules of procedure together with the list of participants — and commits to registering the meetings and issuing high-level summaries, subject to commercial confidentiality.

This places the Taskforce alongside the AI Act's other new institutional layer, the Scientific Panel and Advisory Forum appointed on 1 June 2026: independent expert advice on one side, structured dialogue with the signatory providers on the other, both feeding the AI Office as it prepares to enforce.

What it means for organisations

For providers of GPAI models the message is concrete: the grace period on enforcement closes on 2 August 2026, and adherence to the Code is the principal way to show compliance. The Commission has indicated it will concentrate its supervision of signatories on whether they keep to the Code; for non-signatories the same obligations apply, but they must demonstrate compliance by other means. Providers carrying systemic-risk models — presumed above a training compute of 10^25 FLOPs, or upon designation by the Commission — face the heaviest scrutiny once enforcement begins.

For organisations that merely use GPAI models, the date is a reminder that the upstream layer of the supply chain is about to come under live supervision; the duties when procuring GPAI do not change on 2 August 2026, but the providers behind those models now operate under a Commission that can act. The full sequence of application dates is set out in the timeline of obligations.

Read more: The GPAI regime · Fines and enforcement. Take the scan.

Sources

  1. https://digital-strategy.ec.europa.eu/en/policies/guidelines-gpai-providers
    EC guidelines for GPAI providers: obligations apply from 2-8-2025, Commission enforcement powers incl. fines from 2-8-2026, legacy models comply by 2-8-2027.
  2. https://digital-strategy.ec.europa.eu/en/policies/signatory-taskforce-gpai-code-practice
    EC page on the Signatory Taskforce: chaired by the AI Office, meets at least yearly, published Vademecum with rules of procedure and member list.
  3. https://digital-strategy.ec.europa.eu/en/news/first-meeting-signatory-taskforce-general-purpose-ai-code-practice
    EC news (12-2-2026): constitutive first meeting on 30-1-2026; rules of procedure adopted by consensus.
  4. https://eur-lex.europa.eu/eli/reg/2024/1689/oj
    Regulation (EU) 2024/1689 (AI Act), authentic text; Art. 101 (GPAI fines up to 3% or EUR 15m) and Art. 113 (dates of application).

Share on LinkedIn

Read next

A

The AI Act's expert bodies: the Scientific Panel and the Advisory Forum

On 1 June 2026 the European Commission appointed the two expert bodies the AI Act foresees: a 60-member Scientific Panel of independent experts (Art. 68) advising on general-purpose AI and systemic risk, and a 174-member Advisory Forum (Art. 67) for broad input. Both serve two-year terms.

U

The GPAI Code of Practice: what is in it and who is it for?

The GPAI Code of Practice is a voluntary instrument (Art. 56 AI Act) that lets providers of GPAI models demonstrate compliance with their duties under Arts 53 and 55. Three chapters: transparency, copyright, safety and security.

U

The GPAI regime: what providers of general-purpose AI models must already do

Since 2 August 2025, providers of general-purpose AI models face their own rules, with extra requirements for systemic risk. With the July 2025 Code of Practice, the mandatory training-data template and the AI Office's enforcement powers from 2 August 2026, the regime is complete.

Dirk Baaijen

About this knowledge base

Compiled and maintained by YRproject — programme and project direction at the intersection of digital transformation, AI and regulation. Every factual claim is traceable to its primary source. YRproject is led by Dirk Baaijen About & method →

A project or programme? Work with YRproject →

The monthly briefing

AI regulation in five minutes: what changed, what is coming and what it means. No spam, unsubscribe anytime.

Your address is used for this only and stored on our own servers.