Why this session matters
The fastest way to turn an AI content win into a crisis is to publish something you did not have the rights to, that impersonated a real person, or that you failed to disclose. None of that requires bad intent - it happens by default when a team moves at AI speed without a governance layer. This session gives you the leader's view of the four exposures - IP, likeness, disclosure, and provenance - and the stack that keeps you fast and defensible. You will start a content IP and disclosure checklist you can put in front of your team and your vendors.
The IP and disclosure risks of AI content 6 min live
This is the loud part of the whole course. AI content is not free of consequences just because a tool made it. There are four exposures every content leader has to hold: copyright and training-data questions (who owns the output, and did the inputs infringe), likeness and right-of-publicity (real people, deepfakes, cloned voices), disclosure (the FTC and platforms expect AI content and endorsements to be disclosed), and provenance and authenticity (content credentials and watermarking, increasingly expected). The governance stack in the diagram is how you address all four without stopping the work.
LiveThe four exposures, named3 min▶
Say these out loud with your team so nobody can claim they did not know. Each one has turned a routine content decision into a legal or reputational problem for real organizations. None of them are exotic - they show up in ordinary day-to-day content work.
- Copyright + training data. Who owns AI output is genuinely unsettled, and purely AI-generated work may not be protectable. Separately, the inputs may infringe - feeding in others' copyrighted work carries risk regardless of the tool.
- Likeness + right of publicity. Generating a real person's face or a cloned voice - a public figure, a customer, an employee - without permission is a serious exposure. Deepfakes and AI voices are the sharpest edge here.
- Disclosure. The FTC and platforms expect AI-generated content and AI-assisted endorsements to be disclosed. Undisclosed AI can be treated as deceptive.
- Provenance + authenticity. Content credentials and watermarking are becoming table stakes for showing where content came from - increasingly expected, not optional.
The campaign pulled for a face nobody cleared. A team generated a striking AI portrait for a launch. It resembled a real, identifiable person closely enough to draw a complaint, and the campaign was pulled mid-flight. The image was beautiful and the tool was licensed - but the likeness was never cleared. The base of the stack, rights-clear inputs, was missing, and everything above it fell.
Questions to ask your team
- For our AI content, can we say who owns it and that the inputs were rights-clear?
- Have we ever generated a real person's face or voice without written permission?
- Where are we legally required to disclose AI, and are we actually doing it?
Self-studyRead the stack from the bottom up2 min read▶
The diagram is a stack for a reason: each layer rests on the one below it. Skip the base and everything above is exposed - a perfect disclosure line does not save you if the input infringed. Read it bottom to top.
- Rights-clear inputs - the foundation. No others' IP, no real likeness or voice without permission. Get this wrong and nothing else matters.
- Human edit + fact-check - a named person owns the claims and the voice, and no invented stats or quotes ship.
- Disclosure - flag AI content and endorsements where the FTC or platforms require it.
- Provenance - adopt content credentials and watermarking as they mature, so you can show where content came from.
- Policy + sign-off - a named owner clears it against the policy. Nothing publishes without this top layer.
This is the same governance layer your creators build hands-on in b10 - here you are setting the policy they will run.
Questions to ask your team
- Which layer of this stack is weakest for us right now?
- Who is the named sign-off owner, and do they have the authority to say "not yet"?
The governance map 6 min live
You do not need to be a lawyer to lead this well - you need to know the four issues, what each one means in plain language, and what to check before you publish. The table below is that map. One honest caveat: this area moves fast, so treat the specifics as current-best-understanding and re-verify with counsel before anything high-stakes ships.
Self-studyFour issues, what they mean, what to check2 min read▶
| Issue | What it means + what to check |
|---|---|
| Copyright + ownership of AI content | Ownership is unsettled and purely AI-made work may not be protectable; inputs can infringe. Check: were inputs rights-clear, and do we need human authorship to claim rights? |
| Likeness + voice rights | Real faces and cloned voices carry right-of-publicity exposure. Check: is any real person identifiable, and do we have written permission? |
| FTC disclosure | AI content and AI-assisted endorsements may need clear disclosure. Check: is AI involvement disclosed where required, in plain language the audience sees? |
| Provenance + content credentials | Authenticity signals show where content came from. Check: can we attach content credentials or watermarking, and does the platform expect them? |
Questions to ask your team
- For each of these four issues, what is our actual current practice - not the aspiration?
- When did we last check these rules against something authoritative, and who owns keeping it current?
The governance stack you need 4 min live
Knowing the risks is not governance. Governance is a small set of steps that run every time, owned by named people, on a review cadence. Five pieces make a stack that is strong enough to be safe and light enough that the team will actually use it - the mirror of what your creators operationalize in b10.
LiveThe five pieces of a working stack2 min▶
- An IP-clearance step. Before generating, confirm the inputs are rights-clear and no real likeness or voice is used without permission. This is the base; it is non-negotiable.
- A disclosure policy. A simple, written rule for when and how you disclose AI content and endorsements, in plain language the audience actually sees.
- A provenance practice. Attach content credentials or watermarking where you can, and keep records of what was AI-generated.
- The human edit + fact-check gate. A named person owns the claims and the voice; no invented stats or quotes ship. This ties directly to the anti-slop standard in a3.
- Sign-off + a review cadence. A named owner clears each piece, and the whole policy gets re-verified on a schedule because this space moves.
Questions to ask your team
- Which of these five pieces do we already run, and which are missing?
- Is our governance a checklist attached to publishing, or a document nobody opens?
- When is our next scheduled re-verification, and who owns it?
Run a content IP + disclosure gap-scan ◐ 30-45 min total
A short, honest audit of where you stand against the stack - and a set of questions for the agencies and vendors making content for you. Draft it with your leads, then commit to fixing the biggest gap first.
- Walk the stack bottom-up for your team: rights-clear inputs, human edit + fact-check, disclosure, provenance, policy + sign-off. Mark each green, amber, or red - honestly.
- For every amber and red, name one concrete fix and one owner. Start with the base of the stack; a weak foundation undermines everything above it.
- Write your one-line disclosure rule and decide where the audience will actually see it.
- Name the sign-off owner for each channel, and confirm they have authority to hold a piece back.
- Set the date for your next policy re-verification - this space moves, so a standing cadence matters.
Questions to ask your agencies and vendors
- Whose IP is in this content, and can you show me the inputs were rights-clear?
- Have all real faces and voices been cleared in writing? Show me the permissions.
- Where and how is AI involvement disclosed on what you deliver to us?
- What provenance or content-credential practice do you attach, and who signs off before it ships?
Official sources covered
Taught from public IP, disclosure, and provenance canon. Legal specifics move fast and stay with your counsel; this session gives the leader's map. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · Which best describes the IP and likeness risk in AI content?
AI content is not consequence-free. Who owns it is genuinely unsettled, purely AI-made work may not be protectable, inputs can infringe, and using a real person's likeness or voice without permission is a serious exposure.
2 · Why do you disclose AI content?
Disclosure is a compliance issue, not a style choice. Regulators and platforms expect AI-generated content and AI-assisted endorsements to be flagged clearly, in language the audience actually sees.
3 · What is provenance (content credentials, watermarking) for?
Provenance is about authenticity: attaching signals that show origin and AI involvement. Content credentials and watermarking are increasingly expected, so build the practice as it matures.