Questions about the standard

No. Chain of title is the industry term for the paperwork trail that proves who owns and controls a film. You don't need to understand the full system to use HPF. What you need to do is simple: open the AI Disclosure Form, tick the box that describes your production, sign it, and give it to whoever you're delivering or submitting the film to: a sales agent, a festival, a platform, or a broadcaster.

If you do work with a sales agent, they will usually ask for this form as part of their standard paperwork. If you're submitting directly, attach it to your delivery materials the same way you would attach any other release or rights confirmation. You do not need a lawyer to complete it.

No. The taxonomy and all supporting documentation are published under Creative Commons Attribution 4.0 and are free to use.

Yes. The AI Disclosure Form can be completed at any stage: during development, in production, in post, at point of acquisition, or after release. For films already completed, the producer or rights holder completes the declaration based on their knowledge of the production process. There is no retrospective audit requirement.

The HPF AI Disclosure Form is a producer-signed document. It travels with the film in the same legal package as chain of title, rights clearances, and talent agreements, and can be listed as a named delivery item in any sales agreement, licence, or commissioning contract. No new infrastructure is needed: it attaches to the deal documentation your sales agent, broadcaster, or platform manages.

If your production is going through a completion bond or E&O insurance process, the declaration supports those due diligence enquiries. Your sales agent can advise on placement within your delivery schedule.

Before signing a No AI Used declaration, review every tool used across the full production pipeline, including tools used by individual crew members working independently. Pay particular attention to AI-assist features embedded in software you may already use: Photoshop's generative fill, Adobe Firefly, DaVinci Resolve's AI tools, and AI features in most major audio applications may be active by default without being explicitly selected.

The threshold is any software that uses machine learning, neural networks, or generative models to produce, modify, or optimise content in ways that would otherwise require human creative or technical labour. Standard digital tools without machine learning, such as traditional compositing, digital colour grading, or non-AI audio processing, do not count. If you are uncertain about a specific tool, the full taxonomy includes further guidance, and you can submit an edge case query via the consultation address.

A production receives a single classification based on its highest category of AI involvement. A film where AI was used to generate one background environment, with everything else made by human crew, is still classified as Generative AI: the classification reflects the highest kind of AI use anywhere in the production, regardless of scale. The same applies to a composer hired for a film's main theme whose background cues were AI-generated. the AI originated content in that element that appears in the finished film, so the production is classified accordingly.

The declaration form includes a free-text field for producers to describe which elements involved AI and which tools were used. That detail sits in the supply chain documentation between producer, sales agency, and buyer. What flows to audiences is the classification label, not a per-element breakdown.

Tools that act on existing footage, audio, or images go through the reconstruction test. The question is whether the tool reconstructs content that is already present or fabricates content that is not. Reconstruction restores, cleans, or repairs what a human captured, for example noise reduction, de-flickering, artefact removal, or upscaling that sharpens detail already in the source. The output is the same content in better condition, so it is Assistive AI.

Fabrication invents material that was never captured, such as painting in a background revealed by removing an object, or generating detail with no source. That is Generative AI. Where a tool does both, the fabrication decides the classification, and the vendor's label does not. The taxonomy page sets out the test in full.

HPF applies to any content you commission or acquire, regardless of whether it was produced in-house or by an independent producer. For commissioned content, you can include the HPF declaration as a delivery requirement in your commissioning agreement, alongside existing technical and rights delivery requirements. For acquisitions, confirm the declaration is present in the delivery package and retain it in deal documentation.

There is no obligation to display the classification to audiences. If you choose to surface it, options include content information pages, EPG listings, or broadcast credits. The standard is designed to work within existing commissioning and acquisition workflows without requiring new systems.

HPF is directly relevant to festivals, both as a programming consideration and as a disclosure mechanism for your audiences. You can request the HPF declaration form as part of your submission or acquisition process, in the same way you request technical delivery specifications or rights confirmations. The classification can then be included in programme notes, catalogue copy, or any other audience-facing material at your discretion.

Festivals with eligibility criteria relating to human authorship can adopt the HPF taxonomy directly in their submission guidelines, rather than developing their own definitions. The standard is free to use and adapt under CC BY 4.0. Awards bodies considering HPF-based eligibility criteria for human-authored work are particularly encouraged to make contact.

Verification relies on the producer's written declaration, not technical detection. The sales agency or receiving party exercises reasonable commercial reliance on that declaration, in the same way they rely on a producer's representations about chain of title or clearances. Liability for a false or inaccurate declaration sits with the producer who signed it.

No technical system can reliably detect AI involvement across all production types, and audit-based verification would be prohibitively expensive for independent film. The industry already relies on exactly this kind of signed representation for rights, clearances, and chain of title.

A buyer relying on a declaration should hold the producer's own signed form, not simply a sales agency's confirmation that one exists. The agency's confirmation records that a declaration was made; the signed declaration is what carries the producer's warranty to the party relying on it.

The HPF declaration is a warranty: a signed statement by the producer that the classification is accurate. The receiving party relies on it in good faith; liability for any inaccuracy sits with the producer who signed it.

This is the same mechanism used throughout the industry for rights representations, clearances, and chain of title confirmations. It does not require new legal infrastructure.

Completion bond companies and E&O insurers are increasingly including AI-related questions in their due diligence processes. An HPF declaration produces the record they need.

The producer who signed the declaration is responsible for its accuracy. In practice, remediation means notifying relevant parties, issuing a corrected declaration, and amending the classification in outstanding deal documentation by schedule or side letter. Take legal advice in any reclassification scenario.

The EU AI Act's Article 50 transparency obligations apply from 2 August 2026. They require deployers to disclose deepfakes and AI-generated text published on matters of public interest. For films and other evidently artistic or fictional works, the duty is lighter: the existence of AI-generated or manipulated content should be disclosed in a way that does not hamper enjoyment of the work, for example in the end credits or accompanying notes. In June 2026 the Commission published a voluntary Code of Practice to support compliance, which is explicit that signing it is not conclusive proof of compliance.

HPF sits at a different level. It is a producer-level disclosure about the whole film, carried in the delivery documentation rather than shown to audiences, and it covers all AI use, not only deepfakes. A production that discloses under HPF holds the provenance facts an artistic-work disclosure would draw on, but HPF is not a compliance mechanism, and we make no claim that using it satisfies the AI Act or its Code of Practice. We are watching how the rules settle and will keep the taxonomy aligned where that helps.

The taxonomy is currently maintained by HPF. Proposed amendments can be submitted by email or via the GitHub repository, and all responses received before 31 October 2026 will be reviewed for the v1.0 revision.

MSC is committed to finding a neutral home for the standard. The preferred outcome is an existing standards body adopting and maintaining it; failing that, a coalition or membership organisation so it is governed independently of any one company. There is no fixed date for that transfer, and identifying the right body is part of the work ahead. The taxonomy is published under CC BY 4.0. That licence is irrevocable, so what has been published stays published.

The taxonomy covers the full film and television industry: film, scripted and unscripted television. MSC is introducing it through the independent film sales community because chain of title documentation is well placed to carry it. Studios, broadcasters, streaming platforms, television producers, and awards bodies are all welcome to adopt it and engage with the consultation.

HPF is a disclosure standard, not a certification scheme. Binary "certified human" schemes ask whether a production qualifies as human-made, typically with a third-party audit or verification process. HPF does not operate an audit function and does not issue certifications. A production that used generative AI can still make an accurate HPF disclosure. The classification is a neutral record of how AI was used. It carries no judgement about whether that use was appropriate.

The two can coexist. An HPF declaration does not prevent a production from also pursuing a certification if that scheme's criteria apply.

HPF is a disclosure standard, not a collective bargaining instrument. It does not create or satisfy obligations under any labour agreement, and labour agreements do not create or satisfy obligations under HPF. The two operate independently.

They are also complementary. Guilds and unions across the US, UK, Canada, Australia, and elsewhere have negotiated or are negotiating AI provisions that require productions to document how AI was used. An HPF declaration produces exactly that record. Productions should consult their own counsel and the relevant guild or union agreements for the obligations that apply to them.

No. HPF records how AI was used in a production, not whether that use was appropriate. Questions about labour displacement, fair treatment of workers, and compliance with guild and union agreements are important but separate from what HPF does. Productions should consult the relevant agreements for those obligations.

HPF was built for market transparency, not regulatory compliance. But regulation is moving the same way. Across multiple jurisdictions, AI disclosure requirements for content are in development. A standard that works in practice is more useful than one built from scratch to satisfy a compliance requirement.

An HPF declaration produces the kind of record that content labelling frameworks are starting to require. MSC has engaged directly with that process and MSC's policy submissions are available in the Press and Policy section.