Background

The Tool Disclosure List (TDL) is part of maintaining transparency with contractual and security compliance requirements that may restrict or require prior approval for the use of AI/ML technologies in workflows or pipelines delivering client content.

Beyond compliance, the TDL also serves as a foundation for an ethically run AI pipeline. By formally documenting the tools in use, the studio ensures:

• Clear alignment with client contracts, MPA Content Security Best Practices, and TPN audit requirements.
• Proactive disclosure of tools that require client approval before being used in production.
• Prevention of unauthorized or “shadow AI” use in creative workflows.
• Assurance that AI/ML use respects intellectual property rights, performer rights, and industry ethical standards.
• Reinforcement of client trust by ensuring that all outputs are produced under documented, approved, and transparent conditions.

Introduction to AI/ML Tools

Here we provide an overview of AI-enabled features, AI-assisted tools, Generative AI systems, and Cloud-based AI platforms commonly encountered in visual effects (VFX) and animation (ANM) production workflows.

AI/ML tools vary widely in form, function, and integration:

• Some are embedded features within industry-standard commercial software.
• Others are standalone desktop applications or open-source frameworks.
• Increasingly, tools are delivered as cloud-hosted SaaS platforms that process or generate content offsite.

For clarity and consistency, tools are classified into the following categories:

1. AI-Enabled Features (Embedded)

   • Functions built into licensed software, often using AI/ML under the hood, but considered baseline features of the tool.
   • Example: Adobe After Effects motion tracking, Nuke denoising.

2. AI-Assisted Tools (Enhancement/Assistive)

   • Utilities that automate or accelerate discrete production tasks while requiring human oversight.
   • Example: Topaz AI upscaling, rotoscoping aids.

3. Generative AI Systems

   • Applications that create new expressive content (images, video, audio, text) that did not exist before.
   • Always require client disclosure and approval before use in deliverables.
   • Example: Stable Diffusion, RunwayML (local mode), Adobe Firefly desktop.

4. Cloud AI Platforms

   • SaaS or cloud-hosted services where content/data is uploaded, processed, or generated remotely.
   • Always require explicit disclosure and prior client approval due to data security, IP ownership, and residency concerns.
   • Example: Google Veo, RunwayML (cloud), Pika Labs.

This classification ensures that AI/ML tool use is transparent, consistently documented, and aligned with both client agreements and industry compliance obligations, while also recognizing that not all AI tools carry the same purpose, integration, or risk profile.

Tool Disclosure List (TDL)

Disclaimer: This list is not exhaustive and is subject to periodic updates as new tools emerge and existing technologies evolve. We recommend reviewing the latest version prior to each production cycle to ensure continued compliance and transparency.

If you have any concerns, corrections, or additions, please submit them for review so we can maintain the accuracy and relevance of this resource.

Tool Disclosure List Site

This is a searchable, community-driven resource for tracking AI tool disclosures. The data is sourced from a centralized Google Spreadsheet , which serves as our master document. You can easily search, filter, and export entries to CSV or Markdown (MD) formats for your own use.

If you'd like to contribute to the list, please reach out — we're happy to add collaborators.

How to Read and Use the Tool Disclosure Table

This table helps studios, legal departments, auditors, vendors, and artists maintain transparency around AI-assisted tools in VFX and animation workflows. Each column provides specific information relevant to legal, ethical, and production compliance.

Column	Purpose	Values / Examples
Tool Name	Product or platform name.	Adobe After Effects, SideFX Houdini, RunwayML
License Type	Licensing or distribution type.	Commercial, Open-source, Hybrid
Tool Type	Industry category.	AI-Enabled Features, AI-Assisted Tools, Generative AI Systems, Cloud AI Platforms
Tool Category	Ai Functionality.	AI-Enabled Features, AI-Assisted Tools, Generative AI Systems, Cloud AI Platforms
Hosting Model	Where and how the Tool is run and or installed.	Local / On-Prem, Hybrid ( Local install with Cloud based processing), Cloud (SaaS / Platform)
Risk Level	Quick “traffic light” shorthand showing legal/producer/auditor concern.	Low, Medium, High which is tied to the Output Handling category column. ** See Output Handling and Risk Level Categories selections below.
Output Handling Category	Compliance handling bucket.	Standard Software Features – No Approval Required (Low) Generative Content – Approval Required (High) Assisted, Artist-Controlled – No Approval Required (Medium) Internal Use / Non-Deliverable – No Approval Required (Low) Disclosure Required – Producer Decides (Medium)
Usage Description	Plain-English explanation of how the tool uses AI.	“AI assists with denoising and tracking in compositing tasks.”
Official Website	The tool’s official site or reference link.	https://www.example.com

Output Handling and Risk Level Categories

Category	Approval Status	Risk Level	What it Means	Examples	Why
Standard Software Features	No Approval Required	Low	AI functions embedded in licensed, industry-standard tools.	Adobe After Effects motion tracking, Foundry Nuke denoising, Autodesk Flame cleanup	These are standard features of professional tools, not considered “AI-generated deliverables.”
Generative Content	Approval Required	High	AI that generates new expressive content (images, video, text, audio) that could be used in deliverables.	Adobe Firefly, RunwayML, Stable Diffusion (ComfyUI/Automatic1111), Topaz AI upscaling	This is the risk area Studio Execs/Producers want to control. AI-created content in the finished product.
Assisted, Artist-Controlled	No Approval Required	Low	AI assists, but a human artist directs, reviews, and approves the final result.	Photoshop AI-assisted upscaling, manual Topaz AI review, roto/mask suggestions	Final creative control rests with the artist; output is human-directed.
Internal Use / Non-Deliverable	No Approval Required	Low	AI used for technical or workflow efficiency, not deliverables.	Metadata tagging, mocap cleanup, 3D model topology extraction, simulation acceleration	These uses don’t generate expressive content, only optimize internal processes.
Disclosure Required	Producer/Studio Exec to Decide if Approval Needed	Medium	Gray areas where AI use may impact creative results depending on configuration.	SideFX Houdini ML plug-ins, experimental hybrid workflows	Disclosure ensures transparency; Producer/Execs review and decide if approval is required.

Compliance with MPA/TPN+ and NIST 800-171A-R3

The Tool Disclosure List is a core assessment objective for demonstrating compliance. These policies and procedures should be incorporated into your companies security governance framework (SGF).

MPA/TPN+ Control# OR-5.0 – Organizational Security – AI/ML Security Management

AI/ML Security Management Policy

Assessment Objectives	Details
Policy Tailoring	Document and regularly review AI/ML-specific risks and controls.
Risk Management	Assess risks to datasets, models, applications, network, and supporting systems.
Client Approval	Obtain client approval prior to using AI/ML applications in production workflows.
Data Integrity	Verify sources, integrity, and provenance of datasets before use.
Compliance	Align AI/ML usage with local laws, regulations, contracts, and client policies.
Acceptable Use	Define appropriate usage of AI/ML datasets, integrated with Acceptable Use Policy (OR-1.1).
Sandboxing	Restrict generative AI use to internally managed and sandboxed LLMs.
Training	Provide tailored AI/ML security training by job role.
Tool Disclosure	Maintain a Tool Disclosure List of all AI/ML tools in use, categorized by type, risk level, and approval status.