Can your business license its content to AI?
Video has the clearest buyers so far. Consent from the people in it comes first.
Video has the clearest buyers for business content so far. Broadcasters, producers and distributors license footage through marketplaces such as Troveo and Protege, and Bloomberg reported buyers pay about $1 to $4 a minute.
Before you license anything, check who appears in the footage and who owns the words. Community members, coaching clients and podcast guests rarely agreed to AI use.
In plain words
Data licensing means a company gets to use your content for a while, under rules.
Think of it like this
Think of renting out a room. The buyer stays for a term and follows your house rules. You keep the house.
Licensing your business records instead? Start with the pillar guide on licensing company data to AI companies. This page covers content your business owns: video, audio, posts, courses and sessions.
micro1 may pay us a referral fee if a company we introduce goes ahead with it. How we are paid
Which programs take business content for AI training?
Video. Troveo pays content owners a revenue share on gross earnings. It has not published the percentage (Publish Press). Its content-owners page prefers 1080p or 4K and says processing takes 1 to 2 weeks. Troveo also runs a separate business-data program.
Protege works with broadcasters, producers and distributors. They license content they own outright or control with broad rights. No minimum is stated and the share is undisclosed (Protege).
Calliope Networks said it needed at least 25,000 hours of content (Tubefilter). Protege bought it on 18 Dec 2024 (Protege).
New footage of hands-on work. micro1's video data program asks companies with at least 20 employees to film first-person video of physical work. That is new footage recorded for micro1, so your staff are the people on camera.
Audio. There is no public per-hour rate yet. AIPodcast.io offers back-catalog licensing with a quarterly revenue share and written guest re-clearance. It publishes no rates (AIPodcast).
Text. Pay-per-use is the live model, and it favors sites with traffic. ProRata says it pays publishers 50% of all revenue from its Gist.ai product (News/Media Alliance, WAN-IFRA).
Courses, community content and coaching transcripts. No named buyer pays for these in public reporting so far.
What do buyers pay for video?
AI companies pay about $1 to $4 per minute of video. For ordinary YouTube-style footage it is $1 to $2 (Bloomberg via The Star). Those are buyer prices.
Public sources don't show how much of that price reaches the business that owns the footage. Troveo's revenue share is unpublished, and so is Protege's.
Who owns the words and faces in your content?
Platform terms say what the platform may do with your content. They don't decide what you may license to a third party. That depends on your copyright and on releases from everyone in the content.
| Platform | What the terms say |
|---|---|
| Skool | Content is yours. Skool gets a sublicensable license to provide, develop and improve its services. No AI clause, and the terms don't address admin rights over members' posts |
| Circle | Irrevocable, sublicensable license to process your content and use it, in non-identifying form, to develop and improve Circle's products |
| Kajabi | You keep your content and stay responsible for the consents your students' personal data needs. You may not use MCP data to train AI without Kajabi's written consent |
| Substack | Your license to Substack does not permit it to train third-party AI models |
| YouTube | Google trains Gemini and Veo on YouTube content and channel owners cannot opt out (CNBC) |
The Substack clause limits Substack. As we read it, it says nothing about what you may license from your own posts.
Community members. Members own their posts. As we read the terms, a business that licenses its Skool or Circle archive is licensing something the terms do not hand over. There's no public case of this yet, so the risk is untested.
Coaching clients. Sessions carry confidentiality duties. Tandem Coach reads the ICF's Standard 2.5 on boundaries as meaning clients should know when AI tools touch the coaching relationship.
It suggests disclosure in writing and consent spoken at intake. The standard itself doesn't mention AI (Tandem Coach).
No public market for these transcripts exists yet. Removing names from conversational text is harder than stripping fields (Frankfurt Kurnit).
Podcast guests. Most signed releases before AI licensing existed. AIPodcast re-clears guests in writing and drops episodes where a guest declines or cannot be reached.
People on camera. Content with identifiable people needs signed waivers (AI Optimist). That includes staff who appear in training videos or webinars.
What should you do before licensing anything?
- Inventory what the business owns. List its video, audio, courses and posts by type, and who made each: staff, contractors or guests.
- List the people. Guests, clients, members and anyone on camera or named. This list decides what is licensable before a buyer enters the picture.
- Collect what each person signed. Releases, guest agreements, community terms, client contracts and contractor agreements. Where nothing mentions AI, treat it as an open question for a lawyer.
- Flip the switches you control. YouTube third-party training is off unless you turn it on. Substack's "Tell AI tools not to train" toggle applies only to the AI tools that respect it (Substack Support). Spawning runs a Do Not Train registry (Spawning).
- Ask the contract questions. Revenue share, exclusivity, term, revocation, and whether your files pool into dataset-wide deals you never see. The pillar guide has the full checklist.
- Talk to a lawyer and an accountant. Pick a lawyer who handles media or IP licensing, and see them before you sign. An accountant can explain how any payout is taxed. We are neither.
How do you do this safely?
Run the rights check before you list anything with a buyer. It asks seven questions about whose words and faces sit in your content and what each person signed.
The risk map covers the legal and security failure points for data owners. The buyer directory and deals tracker show who pays what, with dates. Keep your files in your own storage until a signed agreement says who receives them and when they are deleted.
Your business may also hold years of email, CRM or ticket records. The readiness check for business data takes ten minutes. The free fit check comes next.
Do step two above first, today. The list of people in your content sets the limit on everything else.
Apply to micro1's data partnership (paid link)micro1's own data-partnerships pagePaid link: micro1 may pay us a referral fee if a company we introduce goes ahead with it. You pay nothing extra.
Vaultfish helps you run the process. We are not lawyers and this is not legal advice. If you need counsel who knows this area, we can point you to one. Ask for a referral
