A model that takes or produces more than one modality (text, image, audio, video). Each modality is a new channel for personal data, injected instructions and synthetic content that may need marking 1, so guardrails and evals are needed per modality.
- Developed in
- ch. 11, Multimodal models
- Chapters
- ch. 11, AI Defined
- Source
- 1 numbered reference, listed below
Where it is used
The term is not used under this name in running prose; the sections listed under "Developed in" treat it.
Related terms
Sources
- [1] Regulation (EU) 2024/1689 (AI Act), consolidated text as amended by Regulation (EU) 2026/1744 (Digital Omnibus on AI, in force 27 Jul 2026; definitions in Art. 3, incl. 3(1), 3(3) to 3(14), 3(14b), 3(20), 3(22), 3(23), 3(29) to 3(32), 3(49), 3(55) to 3(57), 3(60), 3(61), 3(63), 3(68); Arts. 4, 5, 6 (incl. 6(3) third subparagraph, profiling), 9, 10, 11, 13, 14, 15, 17, 22 to 27 (incl. 26(11)), 40, 41, 43, 47, 48, 50, 53 (incl. 53(1)(c)), 55, 57, 60, 72, 73, 86; Annexes I, III, IV). Publications Office of the EU (EUR-Lex). 2026-07-27. https://eur-lex.europa.eu/eli/reg/2024/1689/2026-07-27/eng (verified: primary)
Definitions of legal terms paraphrase the cited text, which governs. Dated statements are as of .