AI Filmmaking Glossary

California Assembly Bill 1836, which extends post-mortem publicity rights to cover AI-generated digital replicas of deceased performers. Requires consent from the estate for any commercial use of a deceased performer's digital likeness created using AI.

California Assembly Bill 2602 (effective January 1, 2025), which prohibits the use of digital replicas of performers without their informed, specific consent in a contract negotiated by their representative. Applies to voice, likeness, and performance data. Non-compliance can result in contract voidability and damages.

A formal statement describing how AI was used in a production, what tools were employed, what human oversight was applied, and where AI-generated content appears in the final deliverable. Required by an increasing number of film festivals, distributors, and insurance carriers.

A festival policy requiring filmmakers to disclose any AI usage in their submitted work. As of 2026, major festivals including Sundance, TIFF, Cannes, Venice, and Berlin have implemented various levels of AI transparency requirements. Requirements range from simple disclosure checkboxes to detailed per-tool breakdowns.

A formal log documenting every AI tool used in a production, including the tool name, version, purpose, input/output descriptions, human oversight applied, and the resulting gradient level. Required by many festivals and increasingly expected by E&O insurers. AIFilmSafe's Provenance Ledger automates this.

Imperceptible digital markers embedded in AI-generated content that identify it as AI-produced. Distinguished from visible watermarks — AI watermarks are invisible to viewers but detectable by specialized tools. The EU AI Act may require AI watermarking for synthetic media distributed in EU markets.

A verifiable digital badge issued by AIFilmSafe that certifies a production has completed AI compliance documentation. Includes a QR code linking to a public verification page displaying the project's gradient level, compliance status, and certification date. Designed for embedding in festival submissions, press kits, and distribution packages.

An open technical standard for certifying the provenance and authenticity of digital content. Embeds cryptographic metadata in media files documenting their creation history, edits, and AI involvement. Required for EU AI Act Article 50 compliance starting August 2026. Major platforms (Adobe, Microsoft, Google) support C2PA content credentials.

The documented sequence of ownership and rights transfers for creative content. In AI-assisted filmmaking, chain of title must account for which elements are human-authored (copyrightable) and which are AI-generated (potentially uncopyrightable under Thaler v. Perlmutter). AIFilmSafe provides specialized chain-of-title templates that address AI provenance.

The currency used within AIFilmSafe to access AI-powered compliance features. Different actions cost different credit amounts: basic summarizations (5 credits), digital scan notices (15 credits), human-plus audits (20 credits), and E&O export packages (50 credits). Credits are included with subscriptions or available as Project Passes and Producer Packs.

The complete set of documentation a filmmaker needs to demonstrate responsible AI use. Typically includes: AI Usage Register, Performer Consent Records, Chain of Title documentation, Festival Submission Addendum, and the E&O Insurance Preparation Checklist. AIFilmSafe's 5 Master Deliverables map directly to this packet.

A 0–100 score generated by AIFilmSafe's Quick Assessment tool, measuring how well a production has documented its AI usage across five categories: AI Tool Usage, Guild Compliance, Distribution Path, E&O Insurance, and Documentation. Higher scores indicate more thorough compliance documentation.

The consumer-facing name for C2PA provenance metadata embedded in digital files. Content Credentials show viewers the creation history of an image, video, or audio file — including whether AI tools were used. Think of it as a "nutrition label" for digital content.

The process of registering AI-assisted creative works with the U.S. Copyright Office. Requires disclosure of AI-generated elements, "Limitation of Claim" statements for AI portions, and documented human authorship for protectable elements. The Zarya of the Dawn precedent established that selection and arrangement of AI outputs can be registrable.

AI-generated or manipulated media that realistically depicts a person saying or doing something they didn't actually do. In film production, deepfake technology intersects with performer consent laws (AB 2602), SAG-AFTRA's Digital Replica Rider, DGA requirements for directorial approval, and the EU AI Act's Article 50 transparency requirements.

The Directors Guild of America's position on AI in film production. The DGA contract expired June 30, 2026 with negotiations pending. Key stances: directorial approval required for AI-generated visual content, AI cannot replace the director's creative authority, productions must disclose AI use that affects the director's work. DGA members should document all AI-related creative decisions.

An AI-generated reproduction of a real person's voice, likeness, or performance. Regulated by California AB 2602 (living performers) and AB 1836 (deceased), SAG-AFTRA agreements, and the EU AI Act. Creating a digital replica without proper consent exposes productions to legal liability, union grievances, and E&O coverage denial.

A contract addendum required by SAG-AFTRA when a production creates any digital replica of a performer. Specifies the scope of use, compensation, duration, territorial rights, and data security obligations. Must be attached to the performer's deal memo or contract before any scanning or AI processing of their performance.

AIFilmSafe's 5-level classification system (L1–L5) for measuring AI involvement in film production. Ranges from L1 (Ideation Only — minimal risk) to L5 (Primary Generation — high risk). Each level carries specific disclosure requirements, guild implications, and insurance considerations. The scale helps filmmakers communicate their AI usage clearly to festivals, insurers, and distributors.

The AI compliance documentation required by distributors before acquiring a film. Includes chain of title covering AI elements, performer consent records, E&O insurance certificates, and an AI usage summary. Distributors increasingly refuse to acquire films without proper AI documentation — "no documentation, no deal" is becoming the industry norm.

Insurance coverage that protects filmmakers against claims of copyright infringement, defamation, invasion of privacy, and related intellectual property disputes. E&O policies are required by most distributors and festivals. As of 2026, carriers increasingly require AI disclosure documentation before issuing coverage — undisclosed AI use can void your policy.

The European Union's comprehensive AI regulation (entered into force August 1, 2024), with transparency obligations for AI-generated content in media taking effect August 2, 2026. Article 50 requires clear labeling of AI-generated or manipulated content that could be mistaken for authentic. Affects any film distributed in EU territories. Requires C2PA metadata and/or watermarking for compliance.

AIFilmSafe's at-a-glance comparison table showing which festivals require which AI disclosure forms. Tracks disclosure requirements, submission deadlines, and specific AI policies for 20+ major film festivals. Helps filmmakers avoid the guesswork of "which festival wants what."

The complete set of AI disclosure documents required for film festival submissions. Typically includes: AI Usage Addendum, tool-by-tool breakdown with gradient levels, performer consent records (if applicable), human authorship attestation, and training data provenance summary. AIFilmSafe generates this as the Festival 1-Pager export.

AI systems that create new content (images, video, audio, text) based on training data and user prompts. In filmmaking, generative AI tools include text-to-image (Midjourney, DALL-E), text-to-video (Runway, Pika, Kling), text-to-audio (ElevenLabs, Suno), and text-to-text (ChatGPT, Claude). All generative AI usage in production should be documented per SADP.

The lowest level on the Disclosure Gradient Scale. AI was used only for brainstorming, research, or script notes — no AI-generated output appears in the final deliverable. Risk is minimal, and most festivals and insurers require only basic disclosure at this level. Examples: using ChatGPT for script brainstorming or Perplexity for location research.

AI-generated storyboards, mood boards, or pre-visualization materials that are not present in the final cut. Low risk. Examples: Midjourney concept art, DALL-E mood boards, Katalist storyboards.

AI-generated elements that are heavily modified by humans before appearing in the final output. Moderate risk. Human creative control is the key factor. Examples: AI background plates composited and painted over, AI color grading with manual adjustments.

AI outputs used directly in the final deliverable with significant human creative direction. Elevated risk — requires thorough documentation of human oversight. Examples: Runway VFX shots with director oversight, ElevenLabs voice-over with talent consent.

Raw AI generation present in the final deliverable with minimal human modification. Highest risk level — triggers maximum disclosure requirements, potential guild issues, and E&O insurance scrutiny. Examples: AI-generated final VFX shots, synthetic voices as primary characters, AI-generated music in the final mix.

A production standard requiring documented human involvement at every stage of AI-assisted filmmaking — from pre-production through post-production and distribution. HITL defines three levels of human oversight: active participation (human approves each AI output before use), supervisory monitoring (human reviews AI outputs and retains veto power), and audit-based review (human samples and validates AI outputs at scale). Under the EU AI Act, HITL is mandatory for high-risk AI systems. For filmmakers, HITL documentation proves human creative control for copyright (USCO), satisfies guild requirements (SAG-AFTRA, WGA, DGA), strengthens E&O insurance applications, and meets festival disclosure standards. AIFilmSafe's documentation system is built around HITL principles — every AI usage entry captures who reviewed it, what decisions were made, and how human judgment shaped the final output.

The requirement that copyrightable elements of a work must originate from human creative decisions. In AI-assisted filmmaking, this means documenting how humans conceived, directed, selected, arranged, and modified AI outputs. The Triple-A Framework (Aspiration, Augmentation, Assembly) provides a structured approach to proving human authorship.

An AI system trained on vast text data to understand and generate human language. Examples: ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google). In filmmaking, LLMs are used for scriptwriting assistance, research, dialogue generation, and production planning. Typically classified as L1 (Ideation Only) on the Gradient Scale unless output appears in the final deliverable.

The legally required permission from a performer before creating, using, or distributing their digital replica. Under SAG-AFTRA agreements and California AB 2602, consent must be: specific (describing the exact use), informed (the performer understands what they're consenting to), and negotiated through the performer's representative.

A one-time purchase option in AIFilmSafe for filmmakers who need compliance documentation for a single production. Available as Short Film Pass ($49, 500 credits) or Feature Film Pass ($499, 5000 credits). Ideal for filmmakers who don't need an ongoing subscription.

The practice of crafting specific text instructions to guide AI tools toward desired outputs. In the Triple-A Framework, prompt engineering falls under the Augmentation phase — documenting your prompts is key evidence of human creative direction. Well-documented prompt histories strengthen copyright claims and compliance documentation.

AIFilmSafe's project-level record that tracks every AI tool interaction, complete with timestamps, gradient levels, input descriptions, and output classifications. Supports both manual entry and paste-summary intake (auto-detecting 18+ AI tools). Functions as your production's chain of custody for AI-generated content.

AIFilmSafe's free 2-minute compliance check that evaluates your production's AI documentation status and generates a compliance score with specific recommendations for improvement.

AIFilmSafe's structured framework for documenting AI usage throughout the filmmaking process. SADP ensures every AI tool, its purpose, the degree of human oversight, and the resulting output are recorded in a standardized format that satisfies festival requirements, guild agreements, insurance carriers, and copyright law.

The SAG-AFTRA 2026 TV/Theatrical Agreement addressing AI use. Key provisions: informed consent for digital replicas, strike suspension clause (replicas cannot cross picket lines), articulable business reason requirement for scanning, significant additional value principle favoring human performances, uncapped damages for violations, data security requirements for voice prints and facial scans, and enhanced protections for minor performers.

AI tools that generate video footage from text descriptions. Current tools include Runway Gen-3, Pika, Kling, and Luma Dream Machine. Usually classified L3–L5 on the Gradient Scale depending on the degree of human modification. Productions using text-to-video must document the prompts, selection process, and human creative decisions per the Triple-A Framework.

The landmark 2023 U.S. federal court ruling establishing that AI-generated works without human authorship cannot receive copyright protection. The court held that the Copyright Act requires human creative input. This decision shapes the entire landscape of AI-assisted filmmaking — filmmakers must document their human creative contribution to protect their work.

The datasets used to train AI models — which may include copyrighted works. Key concern for filmmakers: tools trained on unlicensed copyrighted material (like Stability AI models, per Andersen v. Stability AI litigation) create downstream liability for productions. E&O insurers increasingly ask about the training data provenance of AI tools used in production.

An evaluation of the legal risk associated with an AI tool's training data. Categories: Low (fully licensed or public domain data), Medium (mixed sources, no active litigation), High (uses scraped copyrighted data), Litigated (tool is party to active training data lawsuits). E&O insurers and festivals increasingly expect this assessment for each AI tool used.

A three-phase model for documenting human creative authorship in AI-assisted filmmaking. Aspiration covers the creative vision and conception. Augmentation covers human direction and prompting of AI tools. Assembly covers the selection, arrangement, modification, and final creative approval of AI outputs. Based on USCO guidance and Thaler v. Perlmutter — essential for proving copyrightable human authorship.

The U.S. Copyright Office's 2023–2025 guidance on registering works containing AI-generated material. Requires applicants to disclose AI usage, identify human-authored elements in "Author Created" fields, and disclaim AI-generated portions in "Limitation of Claim" statements. Following this guidance is essential for protecting copyrights in AI-assisted films.

A contractual guarantee from an AI tool vendor that they will cover legal costs and damages if their tool's training data provenance causes IP claims against your production. Critical for E&O insurance applications — carriers may deny coverage for tools with unresolved training data litigation (e.g., Andersen v. Stability AI). Always check your AI vendor's terms of service.

AI technology that replicates a person's voice to generate new speech. Tools like ElevenLabs, Murf, and HeyGen offer voice cloning. Highly regulated: requires performer consent under SAG-AFTRA agreements and AB 2602. Data security obligations apply to voice prints. Typically classified L4–L5 on the Gradient Scale. E&O insurers often flag voice cloning as a high-risk activity.

The Writers Guild of America's 2026 Minimum Basic Agreement (ratified April 24, 2026, 4-year term through 2030). Key AI provisions: AI cannot be credited as a writer, writers must be informed if AI material is provided to them, studios must disclose training data use involving WGA-covered scripts via a new licensing notification system, and writers retain the right to use AI tools in their own process.

A 2023 U.S. Copyright Office decision on the AI-illustrated comic book "Zarya of the Dawn" by Kris Kashtanova. The USCO ruled that while individual AI-generated images were not copyrightable, the human selection and arrangement of those images could constitute protectable authorship. A critical precedent for filmmakers using AI-generated visual elements.