The AI data market, explained plainly.
Three times a week: what buyers pay, how deals work, and how to stay on the right side of UK data law. Every claim is sourced.
Data brokers and national security: what the UK government is asking
DSIT's 2025 call for evidence asked how brokers keep data away from hostile actors. What it covered, and the checks every broker should run now.
Three months of the data market: what we've learned
A round-up of 40 posts: the prices, the rules and the patterns that matter most for sellers.
Personal data that must never leave: a checklist
Special category data, criminal records, children's data and more: what to exclude before any data deal.
Microsoft paid Taylor & Francis about $10m: academic publishers and AI
Academic publishers were among the first to license archives at scale. Their deals show the value of curated, rights-clear content.
A bankrupt airline's records and a $10m bid: what it means
Spirit Airlines' archive attracted a reported $10m bid in its bankruptcy, subject to court approval. The lessons for UK estates.
Who buys company data? The six types of buyer
From frontier labs to evaluation firms: who pays, what they want, and how they differ.
How to prepare your records for a sale in 30 days
A week-by-week plan to get company records ready for AI licensing buyers: inventory, rights check, de-identification and outreach.
Why years of history can matter more than headcount
In our model, history moves value as much as size does. Here's how the history score builds.
What a fair data licence contract contains
Ten clauses to look for, from use limits to deletion certificates, before you sign anything.
Code repositories: what buyers check before they pay
Ownership, secrets, licences and history: the due-diligence list for selling a codebase to AI developers.
Liquidators: data is an asset, so treat it like one
A practical checklist for office holders: preserve, assess, de-identify and realise, while keeping the file clean.
Accountants: how to talk to clients about their data assets
Your clients may be sitting on an asset they've never valued. A simple way to raise it, and the rules on referral fees.
Photobucket and the per-image price of data: 5 cents to $1
Some data is priced per item. What per-unit deals reveal about how buyers value scale and quality.
The motivated intruder test, step by step
The ICO's benchmark for whether de-identified data is safe: a determined, competent person with no special access.
How a data room works in a licensing deal
Buyers want to see samples before paying. Here's how to show them enough without giving data away.
Why win and loss reasons make CRM data valuable
A pipeline without outcomes is a log. With reasons attached, it's training data for sales agents.
The Data (Use and Access) Act 2025: what changed for UK businesses
The UK's 2025 data law reforms, in brief: recognised legitimate interests, complaints duties and more.
A $1.5bn settlement shows why licensed data is winning
The 2025 authors' settlement with Anthropic put a price on unlicensed training material. Buyers have noticed.
The EU AI Act now asks model makers to summarise their training data
Since August 2025, general-purpose AI providers must publish a summary of training content. Why that favours licensed data.
Support tickets: the most underrated asset in your business
Years of resolved tickets are a near-perfect training set for service agents. What makes them valuable.
What OpenAI pays publishers: the $1m–$5m benchmark
The most-cited price point in AI content licensing, where it came from, and how to use it.
The AI training data market: about $3.2bn and growing over 20% a year
Analysts agree on the direction if not the exact size. What the growth means for sellers.
Generative AI drew $33.9bn of private investment in 2024
The money flowing into AI is why data has a price. The UK figures, and where they go.
Redaction vs replacement: why [PERSON] tags destroy data value
Blanking names out makes text hard to learn from. Consistent stand-ins keep the meaning, and research backs it.
Legitimate interests: the three-part test, explained
The lawful basis most data deals rely on, broken into purpose, necessity and balancing.
Shut-down start-ups are selling their Slack for $10k–$100k
Closure platforms report AI buyers paying for defunct companies' chat, code and email. What sets the price.
PECR and B2B email: who you can contact without consent
Limited companies and LLPs can be emailed without prior consent; sole traders and most partnerships can't. The rules explained.
What is a 100-person company's data worth?
Indicative ranges by company size, anchored on public deals, and why the range is so wide.
The five factors that decide what your data is worth
Scale, history, breadth, rights and buyer fit: how buyers price a company archive, and how to improve each.
Shutterstock made over $100m from AI licensing in a year
Stock libraries and academic publishers were early winners. What their numbers tell other data owners.
Section 171: re-identifying data is a crime in the UK
A little-known part of the Data Protection Act 2018 makes reversing de-identification an offence. Here's how it protects data deals.
Can AI models memorise names? What the research shows
Studies have extracted names and phone numbers word-for-word from trained models. That's why de-identification must happen before transfer.
Exclusive or non-exclusive? The licence term that matters more than price
An exclusive deal can feel bigger today and earn less over time. How to think about exclusivity.
Why a failed company's emails may be its last valuable asset
Stock and equipment sell for pennies. A well-preserved archive can raise real money for creditors, if it's handled right.
UK company insolvencies hit a 30-year high. Their data is still worth something.
Over 25,000 companies went insolvent in England and Wales in 2023. Their records are an overlooked asset for creditors.
The business records AI buyers want most, ranked
Support tickets and code lead; finance records trail. How each system weighs in our valuation model.
News Corp's $250m and what publishers really get paid for AI data
From $1m a year to $250m over five: the spread of publisher deals, and what sets the price.
Pseudonymised isn't anonymised: where the ICO draws the line
Swapping names for codes doesn't take data outside UK GDPR. Here's the identifiability spectrum the ICO uses.
What is company data licensing? A plain-English guide
Licensing your records to AI developers, explained in five steps: what's sold, who buys it, and what you keep.
Reddit's $203m of data deals: what its IPO filing revealed
Reddit's 2024 IPO filing put a number on AI data licensing for the first time. Here's what it means for everyone else.
The public web is running out of training data. Your company's records aren't.
Researchers expect AI developers to use up good public text within years. That's why private business records are becoming valuable.
No posts match that search.