Providers of GPAI models must notify the Commission within two weeks of meeting the systemic-risk threshold
Download: JSON · MarkdownUnder EU AI Act, Article 52(1)
What this label means
- Verified
- — a named reviewer opened the official source and confirmed the record; the date and the reviewer are shown.
- Pending review
- — linked to the official source and in the queue for a reviewer to confirm.
- Source-linked
- — linked to the official source, not yet confirmed by a reviewer.
On this page 9 sections
In brief
Providers of GPAI models must notify the Commission within two weeks of meeting the systemic-risk threshold is a legal requirement under the EU AI Act in the European Union (Article 52(1)). It applies from 2 August 2025. The record carries 2 evidence examples and mappings to 2 frameworks.
- Instrument
- EU AI Act
- Jurisdiction
- European Union
- Nature
- Legal requirement
- Source reference
- Article 52(1)
- Applies from
- 2 August 2025
- Who it binds
- General-purpose AI model provider
- Evidence examples
- 2
- Framework mappings
- ISO/IEC 42001:2023, NIST AI Risk Management Framework 1.0
- Controls that meet it
- 1
- Official source
- Open official text
- Verification
- Verified against the official source 26 Sep 2026
What does it require?
A provider whose general-purpose AI model meets the Article 51(1)(a) high-impact capability condition, which is presumed once cumulative training compute exceeds 10^25 floating-point operations, must notify the Commission without delay and in any event within two weeks of the condition being met or of learning that it will be met. The notification may include arguments that the model nonetheless does not present systemic risk; the Commission decides and keeps a public list of designated models.
Practical action
Track cumulative training compute per model against the threshold and pre-draft the notification and any rebuttal arguments.
Who does it apply to?
Providers of general-purpose AI models approaching or exceeding the compute threshold, including during training.
- Sectors
- Cross-sector / all sectors
- Use cases
- Generative AI and foundation models
Applies from:
Which controls meet this duty?
Satisfies: the control, operated properly, does the work the duty asks for. Supports: it contributes but the duty needs more. Each control page lists every other duty it serves, so work done once can be counted once.
-
satisfiesPolicyHead of AI safety · annualFrontier model safety and security framework
Serves 9 recorded duties · evidence: Published frontier safety framework, Dangerous-capability evaluation report, Threshold notification to an authority
Compute tracking and threshold notification are part of the framework.
Templates that cover this duty
Generated from the records. Free · no account · links emailed to your work address. This duty is cited in each file with its source reference and a link back here.
- AI System Inventory
Register · XLSX and DOCX · v6
- EU AI Act Role and Risk Classifier
Checklist · XLSX · v4
- AI Acceptable Use Policy
Policy · DOCX · v1
- AI Governance Policy and RACI
Policy · DOCX and XLSX · v5
What evidence would a reviewer expect?
| Evidence | Type | Notes |
|---|---|---|
| Training compute tracking record | register | |
| Notification to the Commission | document |
Framework mappings
Original editorial crosswalks. They cite clause numbers only and reproduce no standard text; confidence reflects how direct the mapping is.
See every European Union duty mapped this way →
| Framework | Reference | Note | Confidence |
|---|---|---|---|
| ISO/IEC 42001:2023 | Clause 4.2; Annex A.8.3 | Interested-party requirements and external reporting. | low |
| NIST AI RMF 1.0 | GOVERN 1.1, MAP 1.1 | Legal requirement tracking and system context. | medium |
Frequently asked questions
Is this a legal requirement?
Yes. It is a legal requirement under EU AI Act (Article 52(1)).
Who must comply with this obligation?
Providers of general-purpose AI models approaching or exceeding the compute threshold, including during training.
When does this obligation apply?
It applies from 2 August 2025.
What evidence shows compliance?
A reviewer would expect: Training compute tracking record; Notification to the Commission.
Which frameworks map to this obligation?
It is mapped to ISO/IEC 42001:2023 Clause 4.2; Annex A.8.3, NIST AI Risk Management Framework 1.0 GOVERN 1.1, MAP 1.1. These are editorial crosswalks with a stated confidence, not official mappings.
Which controls satisfy this obligation?
Frontier model safety and security framework.
What should an organisation do in practice?
Track cumulative training compute per model against the threshold and pre-draft the notification and any rebuttal arguments.
Cite this record
AIPolicyTracker (2026). “Providers of GPAI models must notify the Commission within two weeks of meeting the systemic-risk threshold (EU AI Act)” [AIPT-OBL-0042, verified 26 September 2026]. https://aipolicytracker.org/id/AIPT-OBL-0042 (accessed 8 November 2026). Data licensed CC BY 4.0.
Permanent id AIPT-OBL-0042: it never changes or gets reused, and its address keeps working if this page moves.
Cite the official text alongside it: Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence, Official Journal of the European Union, https://eur-lex.europa.eu/eli/reg/2024/1689/oj.
BibTeX
@misc{aipolicytracker2026_providersofgpaimodelsmus,
author = {{AIPolicyTracker}},
title = {{Providers of GPAI models must notify the Commission within two weeks of meeting the systemic-risk threshold (EU AI Act)}},
year = {2026},
howpublished = {\url{https://aipolicytracker.org/obligations/eu-ai-act-art-52-systemic-risk-notification}},
url = {https://aipolicytracker.org/obligations/eu-ai-act-art-52-systemic-risk-notification},
urldate = {2026-11-08},
note = {Data licensed CC BY 4.0. Accessed 8 November 2026}
}
Change history
No change to this record's tracked fields has been recorded since 15 October 2026, when field-level history began.
Similar obligations in other instruments
- Use high-risk AI as instructed, monitor it and inform affected people — EU AI Act, European Union
- Providers must meet the full set of provider duties for high-risk AI — EU AI Act, European Union
- Providers must supply conformity evidence and log access to authorities on request — EU AI Act, European Union
- Non-EU providers must appoint an EU authorised representative for high-risk AI — EU AI Act, European Union
- Deployers, distributors and importers must assume provider duties when they rebrand or substantially modify high-risk AI — EU AI Act, European Union
- Law-enforcement deployers must obtain authorisation for post-remote biometric identification and report annually — EU AI Act, European Union
- Non-EU providers of GPAI models must appoint an EU authorised representative — EU AI Act, European Union
- Frontier developers must protect employees who report catastrophic-risk concerns — California SB 53, California (United States)
Disclaimer: informational only, not legal advice. Verify every claim against the linked official sources and consult a qualified lawyer before acting.