{ "model": { "id": "dttt-integrity", "name": "DTTT AI Content Integrity Model", "version": "0.2", "release_date": "2026-03-16", "release_type": "minor", "status": "pre-release", "description": "A three-axis risk classifier for the ethical dimensions of AI-generated or AI-manipulated content in destination marketing. Produces a four-level Integrity classification (Clear / Caution / High Risk / Not Recommended) and a three-part machine-readable disclosure code covering intervention type, consent basis and audience disclosure standard.", "audience": "external", "url": "https://thinkdigital.travel/ai-transparency" }, "relationship_to_other_models": { "note": "The AI Content Integrity Model is distinct from but complementary to the AI Transparency Model. The Transparency Model records how much AI was involved in producing a deliverable (grade A\u2013E, 1\u201399%). The Content Integrity Model assesses whether that involvement was ethically sound \u2014 covering consent, authenticity and audience disclosure. A piece of content should carry both: a Transparency grade and an Integrity classification with disclosure code.", "transparency_model_version": "1.1", "disclosure_code_format": "[AI-type] \u00b7 [CS-consent] \u00b7 [DC-disclosure]", "disclosure_code_example": "AI-ANIM \u00b7 CS-AI \u00b7 DC-CAP" }, "scope": { "definition": "This model applies to content published externally \u2014 campaign assets, social media posts, destination website content, advertising, event collateral and press materials \u2014 where AI has generated, manipulated or animated visual or persona-based content.", "in_scope": [ "AI-animated still photography (Ken Burns pans, cloud motion, people brought to life)", "AI-generated synthetic imagery depicting landscapes, places or objects", "AI-generated imagery depicting or derived from real people", "AI characters, mascots and fictional synthetic personas used in marketing", "Human-passing AI influencers and spokespeople used in campaigns", "AI-generated text and copy for external publication" ], "out_of_scope": [ "Internal work products, draft materials and research outputs not distributed publicly", "Standard AI-assisted copy and text editing where no real people or imagery are involved", "AI tools used solely for operational or administrative tasks" ], "note": "This model focuses on the ethical dimensions of AI use in content. It does not replace the AI Transparency Model, which records how much AI was involved in producing a deliverable overall." }, "axes": [ { "id": "intervention", "label": "Intervention", "description": "What AI did to or created in the content. This determines baseline ethical exposure and is the first component of the disclosure code.", "codes": [ { "code": "AI-TXT", "name": "Text only", "description": "AI generated or assisted with written content. No imagery, no real people manipulated.", "baseline_risk": "low", "examples": [ "AI-drafted campaign copy", "AI-generated social captions", "AI-written press releases" ] }, { "code": "AI-IMG", "name": "Synthetic imagery", "description": "AI-generated visuals depicting landscapes, places or objects with no identifiable real individuals depicted.", "baseline_risk": "low", "examples": [ "AI-generated destination imagery with no people", "AI-created conceptual visuals", "Synthetic product renders" ] }, { "code": "AI-ENH", "name": "Image enhancement", "description": "AI-applied animation or enhancement of real photography where any people present are unidentifiable \u2014 distant, backs turned or faces obscured.", "baseline_risk": "caution", "examples": [ "Ken Burns pan effect on landscape photography", "AI-animated cloud timelapse over a cityscape", "Lighting enhancement on a crowd scene where individuals are indistinguishable" ] }, { "code": "AI-ANIM", "name": "Animation of real people", "description": "Real photographs or footage manipulated to show identifiable individuals moving, performing actions or appearing in altered settings. The VisitBritain/TripAdvisor scenario of animating a bagpipe player or a walking family falls under this code.", "baseline_risk": "high_risk", "examples": [ "Animating a bagpipe player to appear to be playing", "Making a family in a photograph appear to walk", "Moving facial expressions or gestures on a portrait subject" ] }, { "code": "AI-CHAR", "name": "Character-based synthetic persona", "description": "A clearly fictional AI persona \u2014 creature, mascot or stylised character not intended to pass as human. Audiences can reasonably recognise the character as synthetic.", "baseline_risk": "caution", "examples": [ "Norway's Trollfluencer", "Abu Dhabi's Dinofluencer", "A clearly animated or cartoon-style AI destination mascot" ] }, { "code": "AI-PERS", "name": "Human-passing synthetic persona", "description": "A synthetic human-like persona designed to appear realistic. Includes hyper-realistic AI influencers and spokespeople used in campaigns. Audiences may not immediately recognise the persona as AI-generated without explicit disclosure.", "baseline_risk": "not_recommended", "examples": [ "Germany's Emma AI influencer", "Benidorm's hyper-realistic AI influencer", "Any AI persona where casual scrolling would not reveal its synthetic nature" ] } ] }, { "id": "consent", "label": "Consent", "description": "What authorisation exists for real subjects depicted or represented. This is the central ethical gate. A standard photography release does not cover AI manipulation of a subject's image.", "codes": [ { "code": "CS-NA", "name": "Not applicable", "description": "No real people depicted. Content is fully synthetic or involves only landscapes, objects or unidentifiable figures.", "risk": "low" }, { "code": "CS-ANON", "name": "Anonymous subjects", "description": "People visible in the content but cannot reasonably be identified \u2014 backs turned, distant, faces obscured or in crowd shots where individuals are indistinguishable.", "risk": "low" }, { "code": "CS-AI", "name": "Explicit AI-use consent", "description": "Subjects provided specific written consent for their image or likeness to be used in AI-manipulated or AI-generated content. This must be explicit and separate from a standard photography release \u2014 a standard release does not qualify.", "risk": "low" }, { "code": "CS-STD", "name": "Standard release only", "description": "A standard photography or model release is in place, but no specific consent was obtained for AI manipulation, animation or use in AI-generated contexts. Standard releases typically cover photography and traditional editing only.", "risk": "high_risk" }, { "code": "CS-NONE", "name": "No consent", "description": "Real, identifiable individuals depicted with no model release or consent of any kind. Common in user-generated content or public photography used without rights.", "risk": "not_recommended" }, { "code": "CS-EST", "name": "Estate or deceased", "description": "Deceased or historical individuals. Estate consent not confirmed. Even historical public figures carry significant reputational and legal risk when their likeness is used in AI-animated or AI-generated contexts.", "risk": "not_recommended" } ] }, { "id": "disclosure", "label": "Disclosure", "description": "What the audience was told about the AI nature of the content. Disclosure is necessary but not sufficient \u2014 it cannot resolve a consent failure.", "codes": [ { "code": "DC-PROM", "name": "Prominent in-content disclosure", "description": "A visible label, watermark, badge or on-screen notice embedded in or overlaid on the content itself. Visible to any viewer regardless of whether they read captions, descriptions or metadata.", "standard": "best_practice" }, { "code": "DC-CAP", "name": "Caption or post-level disclosure", "description": "AI use acknowledged in accompanying text \u2014 social caption, post copy, article description or press release note. Visible on-platform to anyone who reads the accompanying copy.", "standard": "acceptable" }, { "code": "DC-META", "name": "Metadata or regulatory minimum", "description": "AI disclosure recorded in file metadata (EXIF, XMP, IPTC) or provided to the platform for compliance purposes. Not visible to the general audience consuming the content.", "standard": "insufficient" }, { "code": "DC-NONE", "name": "No disclosure", "description": "No indication given to the audience that AI was involved in creating or manipulating this content. Not recommended for any visual or persona content regardless of intervention type.", "standard": "not_recommended" } ] } ], "classification": { "description": "The Integrity classification is derived by evaluating the combination of intervention type, consent basis and disclosure standard. It is a risk classifier, not a grade scale. Unlike the A\u2013E models, the classification does not measure degree \u2014 it assesses whether ethical conditions have been met.", "levels": [ { "id": "clear", "label": "Clear", "colour_hex": "#00703c", "text_colour": "#ffffff", "description": "Ethical use confirmed. The combination of intervention type, consent basis and audience disclosure meets the standard for responsible AI use in destination marketing content.", "action": "Proceed with publication. Record this classification alongside the Transparency grade in the Report Card." }, { "id": "caution", "label": "Caution", "colour_hex": "#f4a100", "text_colour": "#0b0c0c", "description": "One or more aspects of consent or disclosure fall below the recommended standard. The content is not automatically unsuitable for publication, but editorial review is advised before proceeding.", "action": "Review the identified risk factors. Consider whether consent gaps can be addressed before publication and whether disclosure can be strengthened. Document the review decision." }, { "id": "high_risk", "label": "High risk", "colour_hex": "#e05206", "text_colour": "#ffffff", "description": "Significant ethical and potentially legal concerns identified. Publishing this content without formal legal and editorial review is not advisable.", "action": "Do not publish without formal review by legal and editorial leadership. Identify whether the consent and disclosure gaps can be resolved. If they cannot, the content should not be published." }, { "id": "not_recommended", "label": "Not recommended", "colour_hex": "#0b0c0c", "text_colour": "#ffffff", "description": "One or more fundamental ethical issues identified that cannot be resolved through disclosure alone. Publishing this content is not recommended.", "action": "Do not publish. The issues identified \u2014 which may include absence of consent for identifiable individuals, use of deceased persons or publishing a human-passing AI persona without any disclosure \u2014 represent failures that a label cannot remedy." } ], "key_principle": "Disclosure cannot fix a consent failure. A Not Recommended classification cannot be resolved by adding a label to the asset. Consent must be obtained \u2014 or the content must not be published." }, "classification_logic": { "not_recommended_triggers": [ "CS-NONE on any content involving identifiable people", "CS-EST on any content involving the deceased", "DC-NONE on any non-text AI visual content", "AI-PERS combined with DC-NONE" ], "high_risk_triggers": [ "AI-ANIM combined with CS-STD (standard release does not cover animation)", "AI-PERS combined with DC-META (metadata disclosure insufficient for human-passing persona)", "CS-STD combined with DC-META or DC-NONE", "AI-ANIM combined with DC-NONE" ], "caution_triggers": [ "AI-ENH combined with DC-NONE", "AI-CHAR combined with DC-NONE", "AI-PERS on any combination (inherent representational risk regardless of consent and disclosure quality)", "CS-STD combined with DC-CAP", "DC-META on any combination not already High Risk" ] }, "disclosure_code_system": { "description": "A structured three-part code for disclosing the nature, consent basis and audience disclosure status of AI-generated or AI-manipulated content. Designed to be attached to any asset \u2014 in metadata, contract deliverables, footers or captions.", "format": "[AI-type] \u00b7 [CS-consent] \u00b7 [DC-disclosure]", "example": "AI-ANIM \u00b7 CS-AI \u00b7 DC-CAP", "embed_formats": [ { "id": "footer", "name": "Document footer", "description": "Colour-coded pill badges with a plain-text description. Suitable for report footers, press release footers and content management systems.", "colours": { "AI-type": "#00857C", "CS-consent": "#005c57", "DC-disclosure": "#003835" } }, { "id": "inline", "name": "Inline badge", "description": "Compact single-line badge with a classification dot. Suitable for social captions, email signatures and slide footers." }, { "id": "metadata", "name": "File metadata", "description": "Plain-text format for embedding in EXIF, XMP or IPTC metadata fields, or for inclusion in asset management system records." } ], "procurement_use": "DMOs can require agencies and vendors to apply the DTTT disclosure code to all delivered AI content assets. Suggested contract language: All AI-generated or AI-manipulated content delivered under this contract must be assessed against the DTTT AI Content Integrity Model and carry a three-part disclosure code in the format [AI-type] \u00b7 [CS-consent] \u00b7 [DC-disclosure]. Assets carrying an Integrity classification of High Risk or Not Recommended require written client approval before delivery." }, "design": { "classification_colours": { "clear": "#00703c", "caution": "#f4a100", "high_risk": "#e05206", "not_recommended": "#0b0c0c" }, "code_pill_colours": { "intervention": "#00857C", "consent": "#005c57", "disclosure": "#003835" } }, "governance": { "licence": "Creative Commons Attribution 4.0 International", "contact": "info@thinkdigital.travel", "committee": true, "versioning_policy": "Pre-release versions track active development before first stable public release. v1.0 will be published once the classification logic has been reviewed by the governance committee and tested across a range of real destination marketing scenarios.", "development_notes": "v0.1 establishes the three-axis structure, the six intervention codes, the six consent codes and the four disclosure codes. The classification logic will be refined at v0.5 following committee review. The human-passing AI persona (AI-PERS) question \u2014 whether a route to Clear exists with sufficiently strong disclosure \u2014 is deferred for committee discussion." }, "changelog": [ { "version": "0.2", "date": "2026-03-16", "type": "minor", "summary": "DC-NONE standard field and description updated from 'unacceptable' to 'not_recommended' to align consistently with the four-level Integrity classification scale.", "changes": [ "DC-NONE standard: unacceptable \u2192 not_recommended", "DC-NONE description: 'Unacceptable for any visual or persona content' \u2192 'Not recommended for any visual or persona content'" ] }, { "version": "0.1", "date": "2026-03-11", "type": "pre-release", "summary": "Initial pre-release of the AI Content Integrity Model. Establishes the three-axis assessment structure (Intervention / Consent / Disclosure), the six intervention codes, the six consent codes, the four disclosure codes and the four-level Integrity classification. Introduces the AI Content Disclosure Code system as a structured three-part machine-readable output. Built in response to emerging industry practices including AI-animated still imagery in destination advertising and hyper-realistic AI influencers in DMO campaigns.", "changes": [ "Model structure defined: three assessment axes \u2014 Intervention, Consent, Disclosure", "Six intervention codes defined: AI-TXT, AI-IMG, AI-ENH, AI-ANIM, AI-CHAR, AI-PERS", "Six consent codes defined: CS-NA, CS-ANON, CS-AI, CS-STD, CS-NONE, CS-EST", "Four disclosure codes defined: DC-PROM, DC-CAP, DC-META, DC-NONE", "Four-level Integrity classification defined: Clear, Caution, High Risk, Not Recommended", "Classification logic documented: triggers for each classification level", "AI Content Disclosure Code system introduced: three-part stackable code in format [AI-type] \u00b7 [CS-consent] \u00b7 [DC-disclosure]", "Three embed formats defined: document footer, inline badge, file metadata", "Procurement clause drafted for DMO supplier contracts", "Key principle established: disclosure cannot resolve a consent failure" ] } ] }