{"id":3071,"date":"2024-01-13T14:55:15","date_gmt":"2024-01-13T14:55:15","guid":{"rendered":"https:\/\/artificialintelligenceact.eu\/annex\/9c\/"},"modified":"2024-08-07T14:32:58","modified_gmt":"2024-08-07T14:32:58","slug":"13","status":"publish","type":"annex","link":"https:\/\/artificialintelligenceact.eu\/annex\/13\/","title":{"rendered":"Annex XIII: Criteria for the Designation of General-Purpose AI models with Systemic Risk Referred to in Article 51"},"content":{"rendered":"<p>For the purpose of determining that a general-purpose AI model has capabilities or an impact equivalent to those set out in <a href=\"https:\/\/artificialintelligenceact.eu\/article\/51\">Article 51<\/a>(1), point (a), the Commission shall take into account the following criteria:<\/p>\n<p>(a) the number of parameters of the model;<\/p>\n<p>(b) the quality or size of the data set, for example measured through tokens;<\/p>\n<p>(c) the amount of computation used for training the model, measured in floating point operations or indicated by a combination of other variables such as estimated cost of training, estimated time required for the training, or estimated energy consumption for the training;<\/p>\n<p>(d) the input and output modalities of the model, such as text to text (large language models), text to image, multi-modality, and the state of the art thresholds for determining high-impact capabilities for each modality, and the specific type of inputs and outputs (e.g. biological sequences);<\/p>\n<p>(e) the benchmarks and evaluations of capabilities of the model, including considering the number of tasks without additional training, adaptability to learn new, distinct tasks, its level of autonomy and scalability, the tools it has access to;<\/p>\n<p>(f) whether it has a high impact on the internal market due to its reach, which shall be presumed when it has been made available to at least 10 000 registered business users established in the Union;<\/p>\n<p>(g) the number of registered end-users.<\/p>\n","protected":false},"template":"","class_list":["post-3071","annex","type-annex","status-publish","hentry"],"meta_box":{"date_of_eif":"","eif_according_to":"","eif_inherit_from":"","item_summary":"<p>Article 51 explains how an AI model is classified as having systemic risk. When it is involved in making this determination, the European Commission will take into account the criteria listed in this Annex.<br \/>\nFactors considered include: the model's complexity, the quality and size of the data used, the computational power needed to train the model, the type of input and output the model uses, the model's performance on various tasks, its level of independence and scalability, and its impact on the market. If the AI model is used by at least 10,000 registered businesses in the EU, it is assumed to have a high market impact.<\/p>\n","item_order":"13","annex-recital_to":["2670"]},"_links":{"self":[{"href":"https:\/\/artificialintelligenceact.eu\/wp-json\/wp\/v2\/annex\/3071","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/artificialintelligenceact.eu\/wp-json\/wp\/v2\/annex"}],"about":[{"href":"https:\/\/artificialintelligenceact.eu\/wp-json\/wp\/v2\/types\/annex"}],"version-history":[{"count":5,"href":"https:\/\/artificialintelligenceact.eu\/wp-json\/wp\/v2\/annex\/3071\/revisions"}],"predecessor-version":[{"id":5202,"href":"https:\/\/artificialintelligenceact.eu\/wp-json\/wp\/v2\/annex\/3071\/revisions\/5202"}],"wp:attachment":[{"href":"https:\/\/artificialintelligenceact.eu\/wp-json\/wp\/v2\/media?parent=3071"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}