Data and AI - Business Opportunities: [email protected]
Joined February 2013
- Tweets11.8K
- Following41
- Followers497
- Likes2.2K
Notice the licensing for AI training is an explicitly separate permission.
Cashmere and Newstex announced a license for AI answers (Nov. 5). AI applications can retrieve licensed news and specialist coverage when answering a user’s question.
Publishers control which content they make available, and model training requires separate permission. That’s a concrete example of why “Can we use your data for AI?” is too vague a question for a license.
Hey, I'll build or check a financial model for you for free if I can pitch you on licensing your data. I built custom f. Models for 10 years for all industries and client sizes.
OneMedNet received a mid-six-figure repeat order for radiology data (Nov. 4). After receiving the first dataset, an AI developer ordered more. OneMedNet attributed the reorder to quality and completeness. The buyer received complete, de-identified imaging studies with the original pixel data and details about how each scan was taken.
Spirit airlines has gone out of business, RIP. but they have a data sale coming up and $GOOGL looks like it will be be the winning bid at around $10M on Nov. 6th.
And yes they are buying it for the purpose of AI training.
Mercor bid $7.5M
Micro1 bid $12.5M (this bid came after the auction and that's why it's higher)
The main value in the records is that it shows how an actual airline operates.
Lawmakers cited approximately 100 million emails and 500 million Microsoft Teams messages in the proposed transaction; the final delivered scope is subject to exclusions and safeguards.
Privacy is the central obstacle. The proposed safeguards include:
- Excluding passenger databases, including specified transaction, refund, onboard-purchase and Wi-Fi-sales systems.
- Having Tonic.ai independently de-identify remaining material, such as corporate emails, before delivery.
- Testing for re-identification risk, certifying the processing to the court, and restricting attempts to reconnect records to individuals.
As you will see in court rulings about this, the big challenge is figuring out what can be transferred and used safely.
Update on Meta's proposed commercial terms fiasco:
They are removing the public preview and said the updates that mentioned how they would use connected website data for AI training will not happen on November 20th. It seems they will not be doing this for the foreseeable future.
Remember this was about Meta being able to scrape any website data that used their platform without paying website owners and use that data to train AI.
Manufacturing owners: when your team solves a quality problem, where does the evidence end up?
A rejected batch can leave a useful trail across production, inspection, and maintenance records.
Consider whether your business can connect:
- The material batch and production conditions
- The inspection result and defect classification
- The investigation into what went wrong
- The adjustment, repair, or rework
- The follow-up measurement showing what changed
Those connections can make the records more useful for understanding production decisions. Depending on a buyer’s needs and your rights to share them, they may also be relevant to AI applications involving quality inspection or process analysis.
Start with one production line or recurring quality issue. Describe the available records and identify the person responsible for each source.
I built a free Business Data Checklist to make that first inventory easier. Choose “Manufacturing & industrial,” mark what you have or need to check, and download your selected list. datahasvalue.ai/p/business-d…
Start your inventory
Keep trade secrets, customer-owned specifications, and confidential supplier information out of your checklist notes. Ownership and permission need separate review before records are shared.
An inventory is not a valuation or an approval to license data. I may earn a referral payment if a company joins through my licensing links and meets its program requirements.
How valuable is your company data? Here is my framework that can be used generally: datahasvalue.ai/2026/11/how-…
What makes your company data more/less valuable?
1. Volume
2. Workflow Complexity
3. Domain Expertise
4. Uniqueness
5. Quality
I'll go more in-depth on those things in the next whiteboard.
Anthropic reportedly struck a licensing deal with the Associated Press. Adweek reported November 4 that AP supplied current election information for Claude under a new content and data agreement. Pricing and duration weren’t disclosed, and neither company confirmed the terms on the record. This illustrates another use for licensed data: providing accurate, current answers. The report does not establish whether AI training rights are included.
If you have a company that went out of business, learn about monetizing the data here: datahasvalue.ai/p/data-licen…
Next thing I'll be talking about is how to get an idea of what is the most valuable data to license and for what purposes.
How sensitive data is handled when you license your data for AI training: datahasvalue.ai/2026/11/how-…
Data Use Fight:
Meta is saying that if you use their services (like promoting your website on a Facebook Business Page) then Meta can use content on your independently owned website to develop AI business tools.
Website owners are essentially saying there should be a paid license for such uses of their website content or permissions defining what purposes are allowed.
The word "training" was removed recently from the terms but it still may be the case that the wording still permits AI training.
Right now everything is paused until this is fleshed out.
Some more on this:
On November 3, a São Paulo court temporarily barred Meta from using participating Brazilian publishers’ content without authorization, including for AI training. It also prohibited retaliation against their accounts. The ruling covers publishers represented by three associations; it is a provisional, limited ruling. ANJ
Separately, the UK competition regulator opened an investigation on October 30 into whether Meta’s commercial terms unfairly exploit businesses. Meta has paused the proposed changes pending consultation. The regulator has not determined that Meta broke the law.
Learn more about the negotiating points around data licensing: datahasvalue.ai/2026/10/ai-d…