AI Pact (voluntary pledges for early implementation of the AI Act)

Summary

A Commission initiative that helps organisations prepare for the AI Act before its rules apply. Under Pillar I, any stakeholder can join a network that shares experience and best practices, including through AI Office webinars. Under Pillar II, companies sign voluntary pledges to adopt an AI governance strategy, map AI systems likely to be high-risk and promote AI literacy among staff, and may add commitments such as identifying risks to health, safety and fundamental rights, ensuring human oversight, marking AI-generated content and labelling deepfakes. Signatories agree to report on their progress twelve months after their commitments are published.

Safety topics

Status notes

The Commission announced over a hundred companies as the first signatories of the pledges on 25 September 2024; the text of the pledges, drafted by the AI Office, is dated September 2024. The pledges are not legally binding and can be signed until the AI Act fully applies. The page's signatory list, updated as new pledges are signed, names 300 signatories at the update of 16 October 2026. The page marks the Pact's first anniversary in December 2025, and says that the AI Office and national authorities assumed enforcement powers on 2 August 2026 and that some high-risk requirements apply only from December 2027.

Sources

Open the official source https://digital-strategy.ec.europa.eu/en/policies/ai-pact

Secondary source: https://digital-strategy.ec.europa.eu/en/news/over-hundred-companies-sign-eu-ai-pact-pledges-drive-trustworthy-and-safe-ai-development

Local copy

Verification

History

1. 2026-10-17notes: First pledges were signed on 25 Sept 2024 by more than 100 companies. The signat → The Commission announced over a hundred companies as the first signatories of th — signatory count updated to the page of 16 October 2026; anniversary and enforcement dates from the page2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
2. 2026-10-17safety_topics: ["risk-based classification", "accountability & liability", "human oversight"] → ["risk-based classification", "human oversight", "transparency & labelling", "de — 'accountability & liability' dropped (no such provision); 'transparency & labelling' and 'deepfakes & synthetic media' added for the optional pledges on marking AI-generated content and labelling deepfakes2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
3. 2026-10-17summary: A Commission initiative encouraging organisations to prepare for and implement A → A Commission initiative that helps organisations prepare for the AI Act before i — fuller account of both pillars and the optional pledges, from the page and the text of the pledges2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)
4. 2026-10-17date_last_updated: 2026-09-17 → 2026-10-16 — the page now shows 'Last update 16 October 2026' (5 October 2026 on the snapshot of 11 Oct); version 1's 17 Sept 2026 is on neither2026-10-17-recheck-europe-1 · Claude Code (claude-opus-5-5)

Related records

EU AI Act European Union · Partially in force

Citation

European Commission (AI Office). AI Pact (voluntary pledges for early implementation of the AI Act). 2024-09-25. https://digital-strategy.ec.europa.eu/en/policies/ai-pact. Status: Published (non-binding, current); verified 2026-10-17 (level A). Local copy archive/EU-04/2026-10-17/official.html retrieved 2026-10-17 16:46:29 UTC, SHA-256 d216fddd4c2bd63fd604a5abd2638bf2a62c37c43201d2731a1143ccb73f457b. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record EU-04.

Improve this record

Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.