Accepting new AI matters

Who owns what your AI tool generates? I draft the answer.

Output ownership across ChatGPT, Claude, Gemini, Midjourney, DALL-E, Stable Diffusion, ElevenLabs, Suno, Copilot, and the rest is governed by 14 different sets of provider terms — each with carve-outs for free vs paid, training-data rights, indemnification, and revenue thresholds. I read those terms, map them to your product, and tell you exactly where the risk lives.

Sergei Tokmakov, Esq. · CA Bar #279869 · Licensed since 2011
01 · Engage me

Three ways to get clarity on your AI rights.

Flat fees. Direct work with me. No hourly surprises unless you specifically request the custom-drafting tier.

Single Provider

AI ToS & Output-Rights Review

One AI tool, one product, one clear answer on what you can and can't do with the outputs.

$349
flat fee · one provider, one product
  • One AI provider's ToS analyzed in full (e.g., ChatGPT Plus, Midjourney Pro, ElevenLabs Creator)
  • Written memo on output ownership, commercial-use rights, and training-data posture
  • Risk callouts tied to your specific use case
  • Tier comparison (free vs paid vs API/Enterprise)
  • 30-minute follow-up call to walk through the memo
  • Turnaround: 3 to 5 business days
Start Review → Wire to PaypalButtons.contractReview349 when /shared/paypal-buttons.js lands.
Custom Engagement

Custom AI License Drafting & Negotiation

Bespoke work: AI clauses for MSAs, model-output license riders, vendor pushback, output-rights schedules.

$240/hr
hourly · scoped estimate before start
  • Drafting AI-specific schedules to MSAs and SOWs
  • Negotiating output-rights clauses with vendors or platforms
  • Building model-output license riders for fine-tuned models or weights
  • Custom platform terms covering AI features in a SaaS product
  • Counterparty negotiation memos and revision rounds
  • Written estimate with ceiling before any hours are billed
02 · Deliverables

What's actually in the $1,500 audit.

A scannable breakdown of every artifact you receive. Each one is written for your specific stack and product, not a template.

01

Provider ToS Map

Every primary AI tool you use, mapped to ownership terms, training-data carve-outs, indemnification, and tier-specific differences. Cited to current provider language as of the engagement date.

02

Risk Memo

Plain-English risk assessment for your specific use case. Active litigation flagged: Bartz v. Anthropic, NYT v. OpenAI, Disney/Universal v. Midjourney, BMG v. Anthropic, and the music industry suits.

03

License Clauses for Your Product

Drafted clauses you can drop into your ToS, MSA, or platform terms. Survives upstream provider terms. Aligned with how you actually deploy AI — embedded feature, agency deliverable, or fine-tuned model.

04

Output Provenance Plan

Recommendations on C2PA watermarking, content credentials, attribution, and disclosures — tied to FTC guidance and platform requirements (Runway, Pika, Canva, Adobe).

05

Revenue-Threshold Triggers

Stable Diffusion's $1M revenue rule, Midjourney's pro-tier requirements, Suno's distribution tiers — flagged so you know when your business outgrows its current AI tooling and needs to switch tiers or providers.

06

Two Strategy Calls

One mid-engagement (after the ToS map is done) to align on scope. One delivery call to walk through the memo and revised clauses.

03 · Triggers

When you need this before you ship.

If any of the scenarios below describe your business, the audit pays for itself the moment a client, vendor, or insurer asks "do you actually have the rights to that?"

Creator / Agency

You generate client deliverables with DALL-E, Midjourney, or Runway

Your client buys a campaign or video. They ask whether your imagery is copyrightable and whether they can sublicense it. The answer depends on which tier you use and whether you added human authorship.

Agency / Studio

You rely on Gemini outputs in client-facing reports

Workspace integration, training opt-out, Vertex AI indemnification, and free-tier carve-outs make this question different from a creator. The audit maps it to your agency's actual deployment.

SaaS / Platform

You're embedding GPT, Claude, or an open model in a feature you sell

Provider terms cascade. Your platform terms have to clear ownership, indemnify your users (or not), and survive the provider's training-data carve-outs. This is the audit's core sweet spot.

Model Builder

You're fine-tuning a model and publishing weights or outputs

Stable Diffusion's RAIL-M license, Llama's commercial restrictions, the $1M threshold, and downstream user obligations all need to be reconciled with how you ship the artifact.

Music / Voice

You're producing music with Suno or voice with ElevenLabs commercially

BMG v. Anthropic and the UMG/Concord lawsuits make audio AI the highest-risk modality right now. Distribution rights, sample clearance, and voice-clone consent all sit in the same memo.

04 · Provider snapshot

How AI provider terms differ on the points that matter.

A four-column snapshot. The full 14-platform comparison covers every variant in detail.

Provider
Output Ownership
Training on Your Data
IP Indemnification
ChatGPT (OpenAI)
User owns
Free/Plus: yes by default
Enterprise only
Claude (Anthropic)
User owns
Consumer: opt-in by default (Aug 2025)
Enterprise / API
Gemini (Google)
User owns
Free: yes · Workspace/Vertex: no
Vertex AI
Midjourney
Paid: yes · Free: CC-BY-NC
All tiers train on prompts
None
Stable Diffusion
User owns
Self-host: none
None · $1M revenue rule (SD 3.x)
GitHub Copilot
Code-dependent
Business: opt-out available
Business / Enterprise

Not sure which package you need?

Send me a one-paragraph description of how you use AI in your business. I'll tell you whether the $349 review is enough or whether the audit makes more sense.

Email me directly
Free triage

Score your AI stack in 90 seconds.

Before you decide between the $349 review and the $1,500 audit, run the AI Output Rights Risk Calculator. Composite score plus a four-part breakdown (ToS exposure, ownership clarity, customer disclosure, cross-jurisdiction) and an estimated dollar exposure tied to your AI revenue.

Try the calculator →
05 · FAQ

Frequently asked questions.

Each answer resolves in one sentence before the expansion. Click any question for the full answer.

Who actually owns the output my AI tool generates?
It depends on the provider's terms and your subscription tier — ownership is one question, copyrightability is a separate one.
Most major providers assign output ownership to the user, but copyright protection is a separate question. Purely AI-generated work generally cannot be copyrighted in the U.S. without substantial human authorship per the Supreme Court's Thaler decision and the Copyright Office's 2025 reports. The audit reviews the actual contractual chain: provider ToS, your tier, your downstream license, and your indemnification posture.
What's included in the $1,500 AI Output Rights Audit?
A ToS map of every primary AI tool you use, drafted license clauses for your product, and a risk memo with citations — in 7 to 10 business days.
I review the Terms of Service for every primary AI tool you use (typically 3 to 6 providers), draft license-clause language for your product or deliverables that survives upstream provider terms, and write a risk memo covering ownership, training data, indemnification gaps, commercial-use restrictions, and active litigation exposure. Two strategy calls are included.
How is the $349 review different from the $1,500 audit?
$349 is one provider, one product. $1,500 is your full AI stack plus the downstream license language for what you ship.
If you use one AI tool for one purpose — "I use Midjourney for client work" — the $349 review is the right tier. If you build on top of multiple AI providers and ship something to your own clients or users (a SaaS feature, an agency deliverable suite, a fine-tuned model), the audit is the right tier because the value is in the cross-provider mapping plus the drafted license clauses.
Can you draft custom AI license clauses for my SaaS or agency contract?
Yes — that's the $240/hour custom drafting tier, with a written estimate and ceiling before any hours are billed.
The $240/hour custom drafting tier is for negotiating AI clauses with vendors, drafting AI-specific schedules to MSAs, building model-output license riders, and similar bespoke work. I scope an estimate with a ceiling before starting. If the work is repetitive enough to fit the audit package, I'll tell you and we'll do that instead.
Do I need this if I just use ChatGPT casually?
No — the free guides at /ai-output-rights/ cover casual use. The audit is for businesses commercializing AI outputs.
The free informational guides at /ai-output-rights/ cover the basics for casual users. The audit is for businesses commercializing AI outputs, embedding AI in a product they sell, or fine-tuning models they distribute. If you're not making money from AI outputs, the free guides are likely enough.
Will the audit cover lawsuits and pending litigation?
Yes — the risk memo flags active litigation that could change provider terms or affect your downstream license.
The risk memo flags active litigation: Bartz v. Anthropic ($1.5B settlement), NYT v. OpenAI, Disney/Universal v. Midjourney, BMG v. Anthropic, the UMG/Concord music suits, and Thomson Reuters v. Ross. I cite the cases and explain the practical exposure for your specific use case — not generic copyright theory.
Are you a California attorney?
Yes — California Bar #279869, licensed since 2011. I work with clients nationwide on contract and licensing matters.
I'm Sergei Tokmakov, licensed in California (CA Bar #279869) since 2011. AI output-rights work is contract and licensing analysis, not state-specific litigation, so I work with clients nationwide on these matters. Engagement is via email and Zoom.

Conversations on this topic

Real questions from the Terms.Law forum where founders, freelancers, and tenants worked through situations like yours.

Get the audit started.

Email me a one-paragraph description of your AI stack and product. I'll send back a scope confirmation and the PayPal invoice within one business day.

Email me to start →

Disclaimer. The information on this page is for informational purposes only and does not constitute legal advice. Reading this page or contacting me does not create an attorney-client relationship. AI-output-rights law is evolving rapidly and the analysis provided in any engagement is current as of the engagement date. Sergei Tokmakov is licensed in California (CA Bar #279869); AI output-rights matters that don't require state-specific litigation are handled nationwide. Past results do not guarantee future outcomes.