Effective July 8, 2026

Human-led, AI-assisted

Sageworx is AI-native. We use AI to move faster and make better work -- never to cut corners. People lead the process end to end. AI assists where it adds quality, speed, or insight. Every deliverable, strategy, and decision carries a human signature, and a human is accountable for it.

That is the whole policy in one line. Everything below is how we live it.

Our principles

1. Human-led, AI-assisted. People are accountable for every deliverable, strategy, and decision. AI is in the room to sharpen the work, not to sign off on it.

2. Right-sized disclosure. We disclose meaningful AI use, and we get opt-in when synthetic media or automated decisioning could shape how an audience perceives or is affected by the work. Where AI only touches minor, non-substantive tasks, we use judgment rather than reflex. When the stakes touch trust, regulation, or authenticity, we default to transparency.

3. Truth in advertising. No AI claims we can’t substantiate. No AI-washing. Influencer and endorsement content carries clear, platform-native disclosure every time.

4. Respect for talent. No cloning of likeness, voice, or personal style without written consent and fair compensation, aligned to current union and industry standards.

5. Privacy by design. We minimize the data we use, protect it in transit and at rest, and never feed confidential client data into third-party tools without written approval.

6. Provenance and authenticity. We attach or preserve Content Credentials (C2PA) where the toolchain supports it, and we label synthetic assets clearly where it doesn’t.

7. Bias reduction. We test for and mitigate harmful bias across datasets, prompts, and outputs before work ships.

8. Security and safety. We assess vendor risk, isolate sensitive workflows, and keep audit trails for material AI use.

9. Compliance-ready. We align to the NIST AI Risk Management Framework, follow FTC advertising and endorsement guidance, and prepare for EU AI Act obligations where they apply.

10. Continuous improvement. We review incidents, update our controls, and retrain our teams as the tools and the rules evolve.

How we use AI, and what we disclose

We work at three levels, and we’re clear internally and with clients about which one a project uses.

Assist -- spelling, grammar, research synthesis, mood boards, alternate cuts, code linting. Behind-the-scenes support that doesn’t change the substance of the work.

Augment -- AI generates drafts or variants that humans review, edit, and shape into the final.

Synthetic -- AI generates net-new images, video, audio, talent, or copy that appears in final assets. We label Synthetic use on deliverables and offer Content Credentials wherever the toolchain supports them.

Approvals

Some uses require sign-off before production, not after.

• Synthetic media in final, public-facing deliverables requires written client approval.
• Talent likeness or voice requires written consent and compensation consistent with current SAG-AFTRA and applicable industry terms.
• Automated decisioning or personalization that could materially affect a person requires client legal or DPO sign-off.

Data and privacy

Our default is no. We do not upload client confidential information or personal data into third-party AI tools without written authorization and applicable data-processing terms. Where use is authorized, we prefer enterprise controls -- tenant isolation, retention limits, access logging. We never use client materials or finished work to train public or shared models. Prompts, inputs, and outputs are treated as client confidential unless agreed otherwise.

Security

We run vendor due diligence on any AI tool that touches client data. We access tools through SSO and MFA where available, hold least-privilege roles, and rotate keys. For significant workstreams, we keep audit logs of model, version, settings, and output.

Quality, bias, and brand safety

Before anything ships, work runs a structured pass: fact-check, IP check, brand-safety check, bias scan, and manual creative review. Sensitive campaigns and synthetic assets get red-teamed before publication. Our checks map to NIST’s generative-AI profile, so the process is audit-ready without slowing the work down.

IP and licensing

We prefer models and tools with commercial-use rights and reasonable indemnity. We keep a clear chain of title for reference material, document stock licenses and prompts, and run image, music, and font checks. We don’t mimic the style of living artists without permission.

Provenance and labeling

We embed Content Credentials where the pipeline supports them and preserve provenance metadata through post. On platforms that don’t yet support provenance, we add visible labels or captions for synthetic assets. Adoption across the industry is uneven -- Content Credentials are the emerging standard we aim to use wherever we can.

Governance

An AI Council -- partners, ops, creative, plus outside legal counsel as needed -- sets policy, approves high-risk uses, and reviews incidents. We maintain a model registry per client and project: tool, version, settings, datasets, approvals, and known limitations. Onboarding and refreshers are mandatory. If an AI-related breach, misuse, or material error occurs, we notify affected clients within agreed timelines.

What makes us different, made safe

Design for disclosure. We build labeling and transparency into the creative, so it adds story value instead of friction.

Content Credentials first. We prefer tools and pipelines that support C2PA, so clients can verify provenance and manage downstream risk.

Union-aware production. Commercials, experiential capture, and game content respect the newest consent and compensation guardrails for digital replicas and synthetic performers.

NIST-aligned QA. Pragmatic checklists mapped to NIST’s framework keep us audit-ready without slowing the work.

Client-facing disclosure language

Short form (credits, captions, footers)

“This project used AI tools under Sageworx’s Human-Led, AI-Assisted policy. All final decisions and edits were made by people.”

Synthetic media label (on or near the asset)

“Contains AI-generated [image/video/audio]. Created and reviewed by Sageworx. Content Credentials available on request.”

Influencer brief line

“Disclose the partnership clearly and upfront using platform-native tools and plain language (for example: ‘Ad’, ‘Paid partnership with [Brand]’). No AI-generated or altered depictions of results or endorsements without prior approval and labeling.”

This is Sageworx’s public AI policy. It is read together with our Privacy Policy. Questions: privacy@sageworx.com.