Code of Practice on Transparency of AI-Generated Content (marcatura ed etichettatura dei contenuti generati o manipolati dall'IA)

Where visual disclosure is possible, Signatories will implement the following design specifications:

a) The icon or equivalent label will comprise, as the main visual element, the capitalised acronym “AI” in the English language (e.g., an ‘AI’ icon), unless use of English is incompatible with applicable national laws on the use of languages in commercial or administrative matters in which case the acronym may be disclosed in the national language. The letters in the acronym will have the same vertical dimension. If resized, the proportions of the letters must be preserved.

b) Additionally, Signatories are encouraged to supplement the acronym in the icon with information regarding whether the deep fake or published text is manipulated or generated with AI in an (interactive) second layer where this is technically implementable, in the icon or next to the icon (e.g., text indicating “modified” or “generated”, as illustrated in Annex 1). Signatories are furthermore encouraged to disclose in the (interactive) second layer what has been modified by the AI system (e.g., text or pictogram describing that a face has been altered).

c) The icon or equivalent label may appear in different sizes depending on the context, as long as Signatories ensure that the disclosure is clear and distinguishable to natural persons. The icon or equivalent label may be expressed in different styles (e.g., contrast ratio, colour, or typography), as long as it remains clear, accessible, and distinguishable, i.e., readable and recognisable to natural persons. An EU icon following the design specifications for visual disclosure is publicly available for everyone to use freely (see Annex 1). Where visual disclosure is not possible (e.g., audio-only content), Signatories will implement the following design specifications:

a) The disclosure will include, at the beginning of the deep fake itself, a short audible disclaimer in plain and simple natural language, either in the same language as the content or in English, disclosing the artificial origin of the audio deep fake in a perceivable manner.

b) Where appropriate, the audio disclaimer will be complemented with information regarding whether it is AI-generated or manipulated. Signatories are furthermore encouraged to disclose what has been modified by the AI system. Alternatively, Signatories may develop and rely on other forms of audible disclosures (e.g., an earcon) ahead of the development of a common EU-wide audio solution by the Task Force, described in Measure 1.3, if the deployment of such an audible disclosure solution is accompanied by appropriate awareness raising measures (e.g., public information campaigns, repeated explanatory notices or disclaimers) to ensure broad understanding of the solution by the natural persons exposed to the deep fake audio. When applying the design specifications of this measure, Signatories will consider the potentially diverse composition of the audience exposed to the content (including diverging levels of AI and digital literacy, language proficiency or general knowledge, and vulnerable user categories such as children and the elderly), and the potentially sensitive nature of the context in which the content is used and disseminated (e.g., in the financial, medical, education or other sensitive sectors). Signatories will ensure accessible disclosure with respect to different content modalities and contexts in accordance with applicable Union law, in particular Directive (EU) 2019/882 (the European Accessibility Act) and Directive (EU) 2016/2102 (the Web Accessibility Directive). This includes but is not limited to application of:

a) audio descriptions or alternative cues for visual disclosure elements;

b) tactile and haptic cues for audio-only content, considering the needs of end-users with hearing impediments (e.g., a vibration alert before audio playing identifying that the audio contains deep fake content);

c) high contrast icons or equivalent labels and screen-reader compatibility, including for end- users with colour vision deficiencies;

d) detectability of the icon or equivalent label by assistive technologies. Signatories are encouraged to implement any available relevant accessibility standard or guideline, including but not limited to the harmonised standard ETSI EN 301 549 “Accessibility requirements for ICT products and services” or the W3C Web Content Accessibility Guidelines 2.1, to the extent the Signatories’ services or products fall within the scope of the applicable accessibility requirements and such standards or guidelines. Signatories that are also providers of online platforms or online search engines, including designated very large online platforms or very large online search engines subject to the obligation to put in place risk mitigation measures under Article 35(1) Digital Services Act (DSA), are encouraged to use the EU icon or another equivalently compliant solution to enable deployers of AI systems to implement their labelling obligation in a manner compliant with this Section of the Code and Article 50(4) AI Act (e.g., by making disclosure solutions available in the platform upload interfaces for deployers).