AI Content Policy

Generative AI has changed creative production faster than the contracts, the case law, and the insurance market can keep up. Agencies are getting hard questions from their producers and general counsel: If we publish work made with an AI tool, whose neck is on the line if a third party sues? Tool vendors are getting equally hard questions in return: Will you indemnify us against everything that could go wrong?

Both questions deserve clear answers. This page is ours.

We wrote it so that you, your producer, and your client's legal team can sit with a single document, understand exactly where Ad Legends stands, and align on who is responsible for what before any creative ships. The contract-form version of this same position lives at /policy/ai-content-addendum, which enterprise customers can attach to their MSA.

1. 1. Ad Legends is a creative tool, not a clearance house.

   We give you AI-assisted creative output. We do not, and cannot, certify that everything you choose to publish is free of every conceivable third-party claim. No reputable AI provider in the world makes that promise — and the ones who promise the most (Adobe Firefly, with its enterprise indemnification) only cover the model output itself, not what the user adds, edits, or combines downstream.

2. 2. You own your output, subject to law.

   Subject to your subscription tier and our Terms, the creative output you generate with Ad Legends is yours to use commercially. U.S. copyright law (Thaler v. Perlmutter and the U.S. Copyright Office's January 2025 guidance) limits federal copyright registration to works with sufficient human authorship — that's a constraint of the law, not of our service, and it applies to every AI creative tool on the market.

3. 3. We pass through, but do not expand, upstream model protections.

   Ad Legends is built on top of leading AI providers — OpenAI, Anthropic, Google, Adobe, Black Forest Labs, fal.ai, ElevenLabs, ByteDance, Replicate, and others. Where those providers extend a copyright shield or indemnification (OpenAI Copyright Shield, Microsoft Customer Copyright Commitment, Adobe Firefly enterprise indemnification, Anthropic commercial protections), we pass that benefit through to you on the same terms — no broader, no narrower. We don't take on the upstream training-data risk ourselves.

4. 4. You are responsible for clearance.

   Likeness rights, voice rights, talent releases, brand and trademark clearance, music synchronization, SAG-AFTRA digital-replica compliance, FTC substantiation, NAD review, deepfake disclosure under state law (including Tennessee's ELVIS Act and analogous statutes) — these are publisher responsibilities, not tool-provider responsibilities. Whoever publishes the creative is the one who must verify before it ships.

5. 5. We will not provide unlimited indemnification.

   No tool vendor in this category does, and we won't pretend to. What we will do is be transparent about which upstream providers we use, what their training-data posture is, what safety filters we apply on top, and how to use our service in a way that maximizes the protections available to you. The 2026 introduction of standardized generative-AI exclusions in commercial-general-liability policies (ISO endorsements CG 40 47 and CG 40 48) makes this honesty more important than ever — neither vendor nor publisher should be relying on insurance that may no longer exist.

6. 6. If you're an agency, you are the publisher in the chain to your client.

   When an agency uses Ad Legends to produce work for its client, the agency is the publisher in that chain. Your obligation runs to your client; ours runs to you. We do not become a party to your downstream agreement, and any indemnification, warranty, or service-level commitment you make to your client is yours alone — not ours. We strongly recommend, consistent with ANA and 4As guidance, that you obtain your client's informed written consent to AI use before delivering AI-assisted work.

A short reference your producer or general counsel can read in under a minute.

• Likeness, voice, and persona

  Does any face, voice, body, signature gesture, or stylistic mannerism in the output resemble an identifiable real person? If yes, do you have written, signed consent from that person (or their estate, if applicable) on file? If not, do not publish.

• Talent and union rules

  Are there performers (real or synthetic) in the work? If yes, have you complied with the SAG-AFTRA Commercials Contract (effective April 2025) digital-replica provisions, including written consent, scope description, and 1.5× scale plus use fees where applicable? Have you cleared any other applicable union, guild, or collective-bargaining agreement?

• Trademarks and brand identity

  Does the output include any third-party trademark, logo, trade dress, packaging, vehicle, or proprietary product configuration? If yes, do you have permission, or does your use clearly fall within an applicable fair-use, nominative-use, or comparative-advertising safe harbor?

• Music and audio

  Does the output include musical composition or sound recording? If yes, do you have synchronization rights, master use rights, and any required performance license? Library tracks: do they extend to AI-assisted creative? Voice clones: do you have rights to the underlying voice?

• Substantive copyrighted works

  Does the output substantially reproduce, evoke, or imitate any specific copyrighted work — a photograph, illustration, character, film clip, or design — that you do not have a license to use? ‘Inspired by’ is not a license.

• Advertising claims

  If the work makes any factual claim about a product or service (efficacy, comparison, superlative, testimonial), do you have substantiation on file consistent with the FTC Endorsement Guides, NAD/CARU standards, and any applicable state or industry-specific regulator? AI-generated text is no exception.

• AI disclosure

  Does any applicable platform, regulator, or your own client require disclosure that AI was used in production? (FTC AI-content guidance, state deepfake statutes including Tennessee's ELVIS Act, EU AI Act Article 50 transparency obligations for synthetic content.) If so, where will the disclosure appear?

• Insurance

  Have you confirmed with your broker that your media-liability, errors-and-omissions, and commercial-general-liability policies cover AI-generated content? Many policies introduced AI exclusions in 2026 (ISO endorsements CG 40 47 and CG 40 48). Do not assume coverage.

• Client consent (agencies only)

  If you are an agency, has your end-client given informed written consent to the use of AI tools (including Ad Legends) in the production of this deliverable, consistent with ANA Media Buying Contract and 4As guidance?

• Sign-off

  Has a named human at your organization reviewed and approved the output for the specific use case it is being published into? Keep that record.

• 4As Content Provenance and AI Disclosure guidance — industry-wide content provenance standards and sample agency policy templates.
• ANA / Venable AI Rider for advertiser-agency contracts — template for the advertiser-agency relationship, including disclosure, ownership, and AI ethics provisions.
• SAG-AFTRA AI bargaining and digital-replica resources — union framework for synthetic performers, voice clones, and digital replicas.
• FTC guidance on AI-related advertising claims — substantiation doctrine applied to AI-generated creative.
• U.S. Copyright Office AI Initiative — January 2025 guidance on copyright eligibility for AI-assisted works.
• Google Generative AI Additional Terms of Service — ownership and developer-specific terms that apply to apps built on Google's generative AI APIs.
• Google Generative AI Prohibited Use Policy — what is and isn't allowed when using Google's generative AI APIs (deceptive content, non-consensual imagery, harmful material).