Responsible AI Platform

Annex XIII AI Act

class="oj-doc-ti" id="d1e38-144-1">ANNESS XIII

Official text

id="anx_XIII"> ANNESS XIII Kriterji għad-deżinjazzjoni ta' mudelli tal-IA bi skop ġenerali b'riskju sistemiku msemmija fl-Artikolu 51 Bl-għan li jiġi ddeterminat li mudell tal-IA bi skop ġenerali jkollu kapaċitajiet jew impatt ekwivalenti għal dawk stipulati fl-Artikolu 51(1), il-punt (a), il-Kummissjoni għandha tqis il-kriterji li ġejjin:

(a) in-numru ta' parametri tal-mudell;

(b) il-kwalità jew id-daqs tas-sett tad-data, pereżempju mkejla permezz ta' tokens;

(c) l-ammont ta' komputazzjoni użata biex jitħarreġ il-mudell, imkejjel f'operazzjonijiet b'punt varjabbli jew indikat b'taħlita ta' varjabbli oħra bħall-kost stmat tat-taħriġ, il-ħin stmat meħtieġ għat-taħriġ, jew il-konsum tal-enerġija stmat għat-taħriġ;

(d) il-modalitajiet tal-input u tal-output tal-mudell, bħal test għal test (mudelli lingwistiċi kbar), test għal immaġni, il-multimodalità, u l-limiti tal-ogħla livell ta' żvilupp tekniku biex jiġu ddeterminati l-kapaċitajiet b'impatt kbir għal kull modalità, u t-tip speċifiku ta' inputs u outputs (eż. sekwenzi bijoloġiċi);

(e) il-parametri referenzjarji u l-evalwazzjonijiet tal-kapaċitajiet tal-mudell, inkluż billi jitqies in-numru ta' kompiti mingħajr taħriġ addizzjonali, l-adattabbiltà biex jitgħallem kompiti ġodda u distinti, il-livell ta' awtonomija u l-iskalabbiltà tiegħu, l-għodod li għandu aċċess għalihom;

(f) jekk ikollux impatt kbir fuq is-suq intern minħabba kemm jilħaq nies, li għandu jkun preżunt meta jkun sar disponibbli għal mill-inqas 10 000 utent kummerċjali rreġistrat stabbilit fl-Unjoni;

(g) in-numru ta' utenti finali rreġistrati.

ELI: http://data.europa.eu/eli/reg/2024/1689/oj ISSN 1977-074X (electronic edition)

////////////////////////$(document).ready(function(){generateTOC(true,'', 'Fuq','false');scrollToCurrentUrlAnchor();});

Source: EUR-Lex, Regulation (EU) 2024/1689 — text reproduced verbatim.

📬 AI Act Weekly

Get the most important AI Act developments in your inbox every week.

Subscribe

Frequently asked questions

What does Annex XIII of the AI Act regulate?

Annex XIII describes the criteria the European Commission uses to determine whether a GPAI model poses systemic risk, under Article 51. These criteria determine whether additional obligations apply.

Which criteria determine if a GPAI model has systemic risk?

Seven criteria: (a) number of parameters, (b) dataset quality/size, (c) amount of compute for training, (d) input/output modalities and state-of-the-art thresholds, (e) benchmarks and evaluations, (f) internal market impact (10,000+ business users), and (g) number of registered end-users.

Is there a threshold for the number of parameters?

Annex XIII lists number of parameters as a criterion (point a), but does not set a fixed threshold. Article 51(1)(a) does mention a threshold of 10^25 FLOP of compute for training as a presumption of systemic risk.

When is a model presumed to have systemic risk?

Article 51(2) provides that a model is presumed to have systemic risk if cumulative compute for training exceeds 10^25 FLOP. The Commission may adjust this threshold through delegated acts.

Does the number of users count for systemic risk?

Yes, point f states that high impact on the internal market is presumed when the model has been made available to at least 10,000 registered business users in the EU. Point g also references the number of end-users.

Are multimodal models more likely to be systemic risk?

Annex XIII, point d, specifically mentions input/output modalities (text-to-text, text-to-image, multimodal) as a criterion, with state-of-the-art thresholds per modality. Multimodal capabilities can therefore contribute to the classification.

Who determines whether a model has systemic risk?

The European Commission determines this based on the criteria in Annex XIII and Article 51. The provider may also self-declare that their model has systemic risk. The 10^25 FLOP threshold creates a legal presumption.

What are the consequences of systemic risk classification?

Providers of GPAI models with systemic risk must comply with additional obligations: model evaluation, adversarial testing, incident reporting to the AI Office, and adequate cybersecurity protection (Article 55).