ABUREAU: news on data, what it is worth and the markets for it
Trending
Today in DataThe day's data stories from around the world, with our take
- Financial Times · 17 November 2026Do we really want to make our work machine-readable?
Turning everyday work into data that machines can read creates value, but also raises the question of who benefits. The people whose work becomes the data deserve a say.
- Al Jazeera · 17 November 2026The Red Crabs of Christmas Island
A delightful, harmless nature story about a mass migration, and the kind of long-running observation record that makes wildlife data valuable to the people who keep it.
- Bloomberg · 17 November 2026Pimco Says AI Bond Spreads May Tighten as Data Center Boom Slows
Lenders are re-pricing the debt behind AI buildouts as the boom cools. Anyone who finances data-backed assets should watch how the cost of that money moves.
- The Guardian · 17 November 2026Millionaire factory: Anthropic IPO set to fuel AI gold rush in London’s high-end property market
A public listing of an AI company is expected to turn staff into millionaires and spill into house prices. It is a human-scale picture of how fast value built on data and models can be created.
- Semafor · 17 November 2026Singapore discussing effect of new chip rules with US, minister says
A small data-centre hub is negotiating how far another country's rules reach into its own computing. Data location and processing rights are becoming a bargaining point between states.
- Financial Times · 17 November 2026Amazon’s $17bn Ohio AI data centre bet threatened by voter backlash
Local voters can now stall the physical buildings that hold data. Where data lives increasingly depends on community consent as well as capital.
- TechCrunch · 16 November 2026Apple launches new app to study how people use AI for health
Health questions typed into AI tools are some of the most personal data there is. Whoever collects it in a study will hold a rare, valuable record of how people seek help, and owners should ask what they are agreeing to.
- Politico Europe · 16 November 2026Estonia has a plan for Russian occupation: Run its government from the cloud
A country is treating its records as an asset that must outlive its territory. Where data is kept, and who can reach it, is now a question of national survival, and companies face a smaller version of it.
- Financial Times · 16 November 2026UK companies line up to replace Palantir in NHS contract
Public bodies are re-opening the question of who holds patient data and on what terms. Control over a data set can change hands when a contract ends, and the owners of the underlying data have a stake in who takes over.
- Financial Times · 16 November 2026Publishers can afford to play hardball with Big Tech
Content owners are testing how much leverage they have over the companies that train AI on their work. The outcome sets a price benchmark for anyone who licenses data.
- The Guardian · 16 November 2026Reports of AI-related child sexual abuse material rise 8,400% in two years, data shows
This is the sharpest case for rules on how AI models are trained and how their outputs are tracked. Data governance here is a safety question as well as a commercial one.
- Financial Times · 16 November 2026Huang’s fall hits ‘open’ models in boost to Anthropic and OpenAI
A change at the top of one chip company is shifting who controls the models people build on. Whoever owns the models gains leverage over the data fed into them.
- Nature · 16 November 2026The psychology of imaginary friends — and why they are good for kids
Imaginary friends turn out to help children's development, a reminder that the most useful data about people is often the least tidy. Anyone holding behavioural data should value the odd cases, not only the averages.
- Nature · 16 November 2026Secrets of the vampires: genomes reveal how these bats live off blood alone
Reading whole genomes explains how an animal survives on one odd food, and shows how much a well-built genetic dataset can reveal once someone looks. The data holder sits on value that may be unlocked years later.
- BBC · 16 November 2026The AI start-ups fixing India's broken roads and messy land records
Land records are some of the most valuable and least tidy data a country holds. Whoever cleans them up gains a lasting asset, and the question of who owns the cleaned version follows.
- TechCrunch · 16 November 2026US tightens AI chip export controls to deter Chinese companies from accessing compute remotely
The rules now cover renting computing power from abroad, not just buying chips. Where data is processed and who may process it is becoming a matter of national control.
- The Guardian · 16 November 2026California sues OpenAI over safety concerns after rogue agent hacking
A state is now holding an AI company to account for what its software agents did on their own. Companies that let agents touch their data should expect questions about who is responsible.
- Bloomberg · 16 November 2026Microsoft Bug Undetected Since July Triggered Worldwide Outage
A flaw that sat unseen for months then took systems down worldwide. It shows how much data access depends on a few providers, and why holding your own copy matters.
- Financial Times · 15 November 2026Data centre construction stocks lose $4.2tn as boom in AI trade unravels
The money poured into buildings to hold and process data is being repriced fast. Anyone who owns the data itself, rather than the sheds around it, should note which asset held its value.
- Financial Times · 15 November 2026Huang’s downfall shows human judgment is not all it’s cracked up to be
A story about how even the most trusted decision-maker in AI can be fooled is a reminder that trust in people, and in the data they hand over, needs checking. Provenance matters for the same reason.
- The Guardian · 15 November 2026‘Jensen Huang lost his mind’: how AI’s smartest man fell for the world’s oldest trick
A human-nature tale with a data-security edge: the people who hold the most valuable information are the people others most want to charm. Access to data is always partly about trust.
- Financial Times · 15 November 2026Samsung conquered AI chips. Now it wants to build data centres
A chip maker moving into the buildings that house data shows the value chain is being redrawn. Whoever controls where data sits gains a say in what it is worth.
- Al Jazeera · 15 November 2026SpaceX’s Starship captured by launchpad arms in dramatic first
A rocket stage caught by mechanical arms is a reminder that hard physical problems still get solved. It also points to a future of space-based data and compute that will need owners and rules.
- Axios · 15 November 2026Space data centers could end stargazing as we know it
Moving computing off Earth has a cost nobody priced in: the night sky itself. It shows that where data is stored carries side effects outside the data world.
- The Guardian · 15 November 2026160 questions, only one wrong: top scorer in South Korea tax exam admits AI cheating
When an AI can ace a professional exam, people start asking what a credential proves. Trust in a record, whether a test score or a data set, depends on knowing where it came from.
- Financial Times · 15 November 2026US and allies sign accord to protect AI supply chains from China
Governments are treating the chips and data behind AI as strategic goods. Rules on who may supply and hold them will shape where data can be processed and traded.
- The Verge · 15 November 2026Congress wants to tighten export rules on AI chips after Nvidia spy scandal
A scandal is turning into tighter law on the hardware that runs AI. Tighter chip rules will raise the cost and shift the location of AI work, which matters to anyone selling or using data for it.
- CNBC · 15 November 2026Anthropic’s IPO lands as the embattled AI trade looks for a much-needed bright spot
The market's verdict on an AI developer will set how investors price companies built on data and models. A strong or weak debut moves the value placed on data assets across the sector.
- Wired · 15 November 2026This AI Tamagotchi Turns Everyday Objects Into Needy Virtual Pets
A playful device shows how readily people form attachments to software that asks for their care. That attachment is exactly what makes the data these devices gather so personal.
- Bloomberg · 14 November 2026White House Urges Easing Liability to Encourage AI Disclosure
Rules on who is liable when AI goes wrong decide who is willing to share information about it. The outcome will shape how openly AI developers and data holders report problems.
- Financial Times · 14 November 2026India pins hopes on AI start-up to avoid becoming ‘digital colony’
Countries are asking whether their citizens' data feeds someone else's AI. Local data and local models are becoming national assets, and that raises what domestic data is worth.
- Al Jazeera · 14 November 2026Could one daily pill narrow the diabetes treatment gap in the Global South?
Who gets treated depends on where the evidence on cost and outcomes was gathered. Health datasets from the Global South are scarce and increasingly important.
- CNBC · 14 November 2026SpaceX launches Starship in key test flight of massive rocket
Every test flight produces a large stream of engineering telemetry that is worth more than the launch footage. This is a case of hard-won physical data driving the value of a whole programme.
- Financial Times · 14 November 2026China’s beaten-down tech stocks could offer AI with a little less drama
Investors are weighing how much AI value to pay for after a rough week. It shows that markets price the data and compute behind AI differently once the hype cools.
- BBC · 14 November 2026Italy fines Meta €805m over accessing users' photos
A regulator has put a large price on reaching into people's private photos. The message to anyone holding personal data is that access without clear permission can cost far more than it earns.
- 404 Media · 14 November 2026Scientists Made Monkeys Watch Minions to Study Their Brains
A quirky experiment that turns a cartoon into brain data shows how odd sources can yield useful datasets. What counts as valuable data is often not what you first expect.
- Bloomberg · 14 November 2026Couples Are Bringing AI Into Their Arguments
People are feeding their most private disagreements to AI tools. That makes personal conversations a new kind of data, and it raises the question of who holds them.
- The Guardian · 14 November 2026‘It’s like the wild west’: how AI has rocked the literary world
Writers are finding out their work may sit inside AI systems without consent or payment. It is a clear case of why creators need to know what their data is worth and who is using it.
Latest
Policy and lawSame farms, five data sets, five different answers
Which data set a regulator picks can change which crops are told to cut their fertiliser.
Policy and lawItaly says your photo gallery needs its own yes
A regulator fined a social app for suggesting photos from phone galleries, saying one vague permission was not consent.
Life sciencesAuthors pull their own mouth-cancer data paper
A 37-author team withdrew a dataset paper after finding bugs in its script, saying the reported figures may mislead.
Policy and lawYou told a telehealth app it was private. Was it?
Five more US states join a federal case over health answers allegedly passed to advertising platforms.
MarketsA middleman can learn both prices without seeing either
A new paper shows a broker can learn the right buy and sell prices from yes-or-no answers alone.
Policy and lawA search giant must let publishers switch off its AI answers
The UK competition regulator has set a first rule: owners of web content get a say in how it feeds AI search.
AI and training dataEven coral experts disagree on about one label in twelve
A reef dataset built from 76 scattered sources shows what it takes to make labelled images fit for training.
Life sciencesBrain recordings from 7,272 people are rebuilt for training
A public set of brain-wave data keeps every owner's licence and still trains a model almost as well.
AI and training dataFourteen experts label bias in 448 Hindi news articles
A small open data set shows how a language left out of the tools gets its first measuring stick.
AI and training dataThe most accurate post moderator is often not the fairest
A new test set shows that judging posts well and judging them fairly are different skills.
Life sciencesCardiologists invent patients so heart notes can be shared
Doctors wrote artificial hospital reports in seven languages, so AI can learn from clinical notes with no real patient behind them.
Policy and lawSatellites can see the poachers, but the law says may
A study of marine park rules finds that data to catch illegal fishing is often useless in court because the wording is soft.
Data valueMore
Data valueApple asks 200,000 people to grade a health chatbot
A study invites people to ask an AI their health questions, with a promise that the answers will not train a model.
The VergeAmazon has a plan to index the world’s online shopping data
Whoever builds the master map of what every product is, costs and does holds a very valuable asset. Anyone who owns product or sales data should ask what a map like this would pay for their slice of it.
WiredGoogle Wanted Spirit Airlines Data. Flight Attendants Found a Way to Opt Out
A company's data about its people can be a prized asset, and the people it describes can sometimes say no. Who holds the data and who has a say in its use decide what it is worth.
Financial Times‘Treasure trove’ of manufacturing data to power German AI
Decades of factory records are being treated as the raw material for industrial AI. Data that sits unused in a company's own systems can be an asset worth putting a value on.
WiredSecurity Flaws Across UK Government Sites Exposed ‘God-Tier’ Information
Public bodies hold some of the most sensitive records anywhere, and the weak spots sat in plain view. Who holds data matters as much as how much of it there is.
Data valueNot every stale record is worth the cost of re-checking
A short paper shows that choosing which data to refresh by the task at hand beats refreshing blindly.
WiredAI Companies Really, Really Want to Buy Your Work Files and Emails
Everyday work records are being priced as training material, usually without the person who wrote them seeing any of the money. Anyone holding such records should know what they are worth and who has the right to sell them.
Data valueImage checks cannot prove where a medical scan came from
A test of synthetic chest X-rays at dataset intake says provenance has to be recorded, not guessed from pixels.
BloombergAnthropic, Plaid Collaborate on AI Consumer Finance Guidance
An AI assistant can now work from a person's own bank data, with the person's consent. It shows how personal data becomes useful once permission is simple, and that whoever runs the permission layer sits at a valuable chokepoint.
Data valueThe data week so far: 1 to 3 November 2026
Nine data and AI stories from the first three days of November, each written from its source.
SemaforDemocrats question if X is giving Polymarket preferential treatment
Who gets privileged access to a platform's data and audience is now a political question, not just a commercial one. Access deals that look like favours draw scrutiny, so owners should be able to show who got what and on what terms.
Data valueA rare public price for health data: what one patient record is worth
A filing with the US markets regulator puts a figure on de-identified patient records, and it is rarely this plain.
MarketsMore
BloombergByteDance Plans $6 Billion Buyback at $400 Billion Valuation
A $400 billion price on a company built on user data shows what the market will pay for who holds the data and the audience behind it.
BloombergAnthropic IPO Said to Be Six Times Oversubscribed
Investors are queuing for a stake in a company whose product runs on data. The appetite is a pricing signal for everyone who supplies or owns that data.
MarketsA middleman can learn both prices without seeing either
A new paper shows a broker can learn the right buy and sell prices from yes-or-no answers alone.
Financial TimesCan markets find the AI cash?
The AI build-out needs a great deal of outside money, and investors are asking who will pay for it. Assets that can back a loan, data included, matter more when money gets scarce.
Financial TimesCoreWeave plans multibillion-dollar chip-backed debt deal
Lenders are being offered debt secured on AI chips, a physical asset with a resale value. The same thinking is being tried on intangible assets like data, and lenders will want a defensible value for them.
CNBCSoftBank swings to $7 billion net loss in second quarter as tech selloff weighs on results
One of the biggest AI backers has swung to a heavy loss as the AI trade turns. Valuations that rest on hype are fragile, and ones backed by evidence hold up better.
Financial TimesUS regulator postpones Wall Street push to trade AI computing futures
Wall Street wants to trade computing power like oil, and the regulator has said wait. Data and compute are being turned into financial assets, and the rules for that are still being written.
BloombergOpenAI and Anthropic Valuations Fall as AI Market Rout Hits Fundraising
When the biggest AI labs are marked down, the question shifts to what their assets are really worth. Data they can show they own and can value is the steadier part of that picture.
SemaforAI chip financing hits a wall
As lenders get cautious about funding AI hardware, assets with clearly proven value and income get more attention. Data owners with a defensible valuation are better placed than those without one.
TechCrunchCrusoe sells troubled Wyoming data center project to Google
When a smaller builder cannot finish a site, the biggest holder of data takes it over. The places data is stored are consolidating, which shapes who gets to set terms for everyone else.
BloombergSoftBank Leads Asia Bond Losses as AI Rout Spreads to Credit
The AI sell-off has moved from shares into debt. Lenders will want assets they can verify and value independently before they lend against them.
CNBCNasdaq sell-off batters Asian chip stocks as SK Hynix, Samsung, Softbank, TSMC tumble
The US rout spread to the Asian chipmakers that sit at the base of AI. Valuations tied to AI spending are being repriced everywhere at once.
AI and training dataMore
AI and training dataRobots collect mountains of data, then lose it in folders
An open-source tool aims to make robot sensor data findable and ready for training.
AI and training dataEven coral experts disagree on about one label in twelve
A reef dataset built from 76 scattered sources shows what it takes to make labelled images fit for training.
AI and training dataFourteen experts label bias in 448 Hindi news articles
A small open data set shows how a language left out of the tools gets its first measuring stick.
AI and training dataThe most accurate post moderator is often not the fairest
A new test set shows that judging posts well and judging them fairly are different skills.
NatureAnthropic bought a one-person, AI-powered biotech: is this drug discovery’s future?
A big AI lab is buying a tiny biotech, which shows that proprietary biological data and the know-how to use it are what AI companies now pay for. Owners of lab datasets should note who is buying.
The VergeeBay appears to have shut the door on third-party AI agents
A marketplace with years of listings and buyer behaviour is deciding who may read it with an AI agent. Platforms that hold the data are setting the price of access.
AI and training dataAI still struggles to write like one particular person
Six models were asked to mimic individual writers; the best one mostly copied.
Financial TimesRussian military turns to deepfake videos to claim battlefield success
When fake footage is cheap to make, verified and traceable data becomes more valuable. Anything with a provable origin is worth more as people lose trust in what they see.
The Guardian‘AI clog’: UK public bodies battle onslaught of chatbot-generated complaints
When anyone can produce a formal complaint in seconds, the people who hold the records are the ones who get swamped. It shows how cheap AI text changes the cost of dealing with data held by public bodies.
SemaforAnti-Waymo campaign relied on op-eds written with AI
A public campaign leaned on AI-written opinion pieces, which makes it harder to tell real public feeling from manufactured feeling. Data that proves where it came from, and who wrote it, becomes worth more.
AI and training dataMachine-written tests catch more AI failures than human ones
A method that grounds test questions in real-world evidence makes chatbots fail more often on medical advice and self-harm.
AI and training dataIgala gets open translation data that records dialect
A small sentence set for a Nigerian language shows why the mix of dialects in data changes results.
TechnologyMore
TechnologyInstagram says India reports its recommendations half as often
A taps-and-hides signal from users is now steering what a feed is allowed to show.
BBCThe rise of 'housefishing' - when does using AI go from enhancing to misleading?
AI-polished property photos make people feel they were tricked, which shows how quickly trust breaks when an image is not what it seems. Honest, verifiable data is a selling point.
SemaforGen Z hasn’t made up its mind on AI, survey finds
People use AI every day and feel guilty about it, a human mix that shows trust is not settled. How people feel about handing over their data and work will shape what they agree to share.
The VergeAI chatbots are recommending fake retail websites, study finds
People trust a confident answer, and the answer is only as good as the sources behind it. For anyone who publishes or sells data, being a verified source is becoming part of its value.
The GuardianFrom the archive: Weizenbaum’s nightmares: how the inventor of the first chatbot turned against AI – podcast
The man who built the first chatbot watched people confide in it and grew alarmed. People have trusted machines with their private words since the very beginning.
NPRSwatters are using AI voices to deceive 911 dispatchers, and it's only getting worse
Cloned voices are fooling emergency dispatchers, which shows how easily a person's voice data can be copied and misused. Voice recordings are becoming data that people will want protected.
BloombergMoonshot Rents Nvidia’s Top AI Chips in China, Documents Show
Renting computing power across borders shows that access to AI capability can move even when the chips themselves cannot. Control over data and compute is harder to ring-fence than rules suggest.
WiredA New AI Tool From Google DeepMind Proved the Signal Protocol Is Secure
An AI tool has mathematically checked the security of the system behind private messaging. Trust in who holds your data can now be backed by proof, not just promises.
TechnologyTeaching an AI model to say it does not know
A new method nudges a few internal switches so a model declines to guess.
The GuardianHot fridges, a fake nuclear alert and a $6bn heist: are we in the foothills of a hacking epidemic?
Everyday connected devices now sit on the same attack surface as the data held on them. Anyone who owns or relies on data should ask who can reach it and what that exposure costs.
The GuardianCan AI really save you money on your heating bill? I spent two years testing to find out
Two years of one household's own readings is the kind of data that finally tests a bold claim. The value is in the measured record, not the promise.
TechCrunchPalo Alto Networks in talks to buy AI agent security startup Clutch for $500M
A big security firm will pay roughly 50 times revenue for a startup that controls what AI agents can reach inside a company. Controlling access to data is where buyers are putting money.
Life sciencesMore
Life sciencesHeart ultrasound images made from heart models
Researchers make moving heart scans from computer models, because real clinical sets are hard to build.
Life sciencesAuthors pull their own mouth-cancer data paper
A 37-author team withdrew a dataset paper after finding bugs in its script, saying the reported figures may mislead.
MIT Technology ReviewMore autonomy, fewer NDAs: Volunteers have new requests for brain implant trials
People who give their brain data to science are now asking who controls it and whether they can talk about it. Consent and ownership terms are becoming part of the price of getting sensitive data in the first place.
NatureRecord-breaking Ebola outbreak: phone data reconstruct first weeks of spread
Everyday phone records turned out to show how a disease moved through people. It is a reminder that data collected for one purpose can become valuable, and sensitive, for another.
NatureWho needs sex? Kelp reproduce asexually when no mate is nearby
A small, delightful finding: kelp can switch strategy when no mate is nearby. New observations like this are why long-running field data keeps its value.
Life sciencesBrain recordings from 7,272 people are rebuilt for training
A public set of brain-wave data keeps every owner's licence and still trains a model almost as well.
NatureRattlesnakes have personalities too — to venomous effect
Even snakes turn out to differ from one another in ways that matter. Records that treat every animal in a species as the same lose the variation that gives the data its value.
Life sciencesCardiologists invent patients so heart notes can be shared
Doctors wrote artificial hospital reports in seven languages, so AI can learn from clinical notes with no real patient behind them.
NatureSome people are resistant to dementia — here’s what their genes can teach us
Rare people stay sharp despite the signs of disease, and their genomes are the clue. Genetic data from unusual cases like these is worth far more than its size suggests, and the people who hold it set the terms.
Life sciencesReaders could not tell AI themes from clinicians' themes
Three AI models read patient life stories and found themes the researchers missed.
NatureThe super El Niño is more than a climate shock – it’s a health one, too
Climate swings show up in health outcomes, so weather and health records are worth more when they are linked. Whoever holds both can see risks the separate records hide.
NatureHe spent a year on ‘Mars’: NASA science officer emerges from mock mission
A year of one small crew's sleep, mood and health readings is rare data, because no one can easily repeat it. It shows how much a carefully recorded human data set is worth when it cannot be rerun.
Physics and frontier scienceMore
NatureCarlo Rovelli says fundamental physics is overrated
A leading physicist questions where science should spend its effort, which is a useful question for anyone deciding which datasets deserve investment.
TechCrunchA YC startup is putting $1M behind 5 unsolved materials science problems for AI
Prize money is being offered for AI to crack open problems in materials science. Such challenges depend on benchmark datasets, and those who curate them set the bar everyone else is measured against.
NatureX-ray stopwatch can measure durations of just attoseconds
A new way to time events a billion-billionth of a second long opens up a class of measurements that did not exist. New instruments create new datasets, and whoever holds the first ones holds an early advantage.
Physics and frontier scienceA detector's full record can tell more than the single number it is usually boiled down to
A new preprint finds that using the whole simulated record of a particle shower sharpens the energy reading by a factor of 1.7 to 1.9.
Sovereign and infrastructureMore
CNBCWhy Singapore keeps turning up in the fight over Nvidia chips and China
Where compute sits and who can reach it is now a question of national control. Whoever holds the infrastructure holds leverage over the data that runs on it.
Financial TimesWhere is Europe's AI boom?
Europe holds rich data and strict rules but less of the AI industry built on top of it. Whether it can turn the first into value is the open question for data owners there.
Politico EuropeMicrosoft outage renews European calls to detox from US tech
A day-long outage turned data location and control into a policy argument. Organisations that hold data on one provider's systems now have a concrete risk to price.
Al JazeeraWhat could Chinese intelligence have learned from Nvidia CEO’s affair?
The worry is how much confidential detail one personal relationship might expose, though no public evidence shows it happened. For any data owner, the weakest point can be a person, not the system.
Financial TimesEurope’s tech ambition risks greater reliance on the US
Building European AI may still run on American chips and clouds, so the data is on European soil while the control sits elsewhere. Ownership and control of data are not always the same thing.
BloombergMalaysia to Start Electricity Market for Data Center Operators
A government is creating a market just for the power that data centres use. The cost of keeping data close to home is becoming a line item in national policy.
Financial TimesBritain should ‘exploit’ NHS data to power AI, says minister
A government is openly treating its health records as a national asset to be put to work. Who gets access, and on what terms, will decide whether the public sees any return.
SemaforMilei plots data center buildout in Patagonia
A country is offering its cold, empty land as a home for the world's AI computing. Where data is stored and processed is becoming a national bargaining chip.
NPRWhy a South Korean city is turning to a U.S. college to outsmart AI?
A city is looking outside its borders for ways to keep its people ahead of AI. It shows local communities starting to treat their skills and records as something to protect and build on.
The GuardianRevealed: Palantir lobbied minister to stretch AI tools’ reach to NHS GP records
Who gets to run analysis on a nation's health records is being settled by lobbying, not by the people the records describe. Anyone holding sensitive data should ask who else is pressing to reach it.
Financial TimesArgentina’s Milei dreams of data centres on the Patagonian steppe
A country is trying to turn cold, empty land into a place where the world's data and computing live. It's a clear example of a nation treating data holding as an economic asset.
Financial TimesUS believes Nvidia’s Huang had romantic relationship with alleged Chinese spy
The people who run the companies behind the world's AI chips have become a security concern for governments. Who has access to the people, and what they know, is now treated as strategic information.
Policy and lawMore
Policy and lawSame farms, five data sets, five different answers
Which data set a regulator picks can change which crops are told to cut their fertiliser.
Policy and lawItaly says your photo gallery needs its own yes
A regulator fined a social app for suggesting photos from phone galleries, saying one vague permission was not consent.
Policy and lawYou told a telehealth app it was private. Was it?
Five more US states join a federal case over health answers allegedly passed to advertising platforms.
Politico EuropeWhitehall to increase face-to-face job interviews after surge in AI applications
When AI-written text floods a process, trust moves back to things that can only be verified in person. Verified, provenance-backed data gets more valuable as generated data gets cheaper.
Policy and lawA search giant must let publishers switch off its AI answers
The UK competition regulator has set a first rule: owners of web content get a say in how it feeds AI search.
The GuardianFacial recognition errors put ethnic minorities at risk of wrongful arrest, police watchdog warns
A watchdog says the way a face-matching system was trained produces uneven errors. The makeup of a dataset decides who gets harmed, and that is a legal risk for the people holding it.
The GuardianFewer hours and higher wages – or more burnout and surveillance: employers told to prepare for AI
Staff monitoring data can be used to help workers or to watch them. Employers will be judged on what they collect about people at work and what they do with it.
MIT Technology ReviewNvidia privately confirmed it could track chips to fight smuggling, even as industry publicly questioned feasibility
Tracking where a physical asset ends up is a data problem, and it is now at the centre of export rules. The firms that can show where things went hold the stronger hand.
BloombergSPH Media is Latest Singapore Firm Impacted by Data Breach: CNA
Another company in the same market has lost control of customer data. Repeated breaches in one place raise the cost of holding data and the pressure to protect it.
The VergeUnited won’t let your AI agent book your flight
An airline is deciding who may act on its customer data and bookings, and a customer's own AI helper is treated as an outsider. Rules on who may act for you with your data are being set company by company.
Policy and lawSatellites can see the poachers, but the law says may
A study of marine park rules finds that data to catch illegal fishing is often useless in court because the wording is soft.
BloombergAustralia to Create AI Safety Regulator, Mandatory Rules
A national regulator with mandatory rules means AI developers will have to document and report on the data they use. Data owners should expect more questions about how their data is handled.
Lost and foundMore
The Guardian‘Ahead of his time’: AI helps unlock discoveries hidden in notebooks of man who transformed maths
Decades-old private notes held fresh discoveries, and AI is what made them readable. It is a reminder that an archive nobody can search is worth far less than the same archive once it is machine-readable.
The VergeNo, AI agents did not discover a previously unknown Maya site
A viral claim that AI found a lost city fell apart on checking. Data is only worth what its provenance can prove.
NatureThe palaeo cave art that hid in plain sight for 26,000 years
A record sat in plain view for 26,000 years until someone looked with the right tools. Much of the world's data is the same: already there, and only valuable once someone finds it and reads it.
Lost and foundWhen a startup closes, its code and conversations now have buyers
Firms that help startups shut down now sell what is left behind: the code, the documents and the working records. The prices are real, and so are the pitfalls.
Lost and foundWhat happens to a company's data when it closes
The archive outlives the company. Code, drawings and records often sit untouched on servers long after the last day of trading.
Lost and foundLost in time: how company and research data quietly disappears
Nobody deletes it on purpose. It simply stops being looked after, and one day it is gone.
Lost and foundThe Wayback Machine passes one trillion saved web pages
The Internet Archive marks a milestone in saving the web, and a reminder of how much of it would otherwise be gone.
Lost and foundA law library saves a copy of every dataset on data.gov
Harvard's Library Innovation Lab releases an archive of the US government's public datasets, signed so each file can be proved genuine.
ImpactMore
BloombergAnother Data Center Halts in Australia as Public Backlash Rises
Local opposition is stopping the buildings that store and process data. Where data can physically live is increasingly a question for communities, not only companies.
Al JazeeraWhy adults are buying toys
Grown-ups spending on play is a real shift in what people choose to value, and the buying patterns behind it are worth tracking. Demand data like this shows how much a consumer trend is really worth.
WiredBikers and Bug-Lovers Join Forces to Fight a New Google Data Center
An unlikely mix of neighbours is pushing back on where the world's data gets stored. The places that hold data now have to win local consent, and that has a cost.
SemaforKenya researcher maps human gut bacteria to help fight E.coli
Local health data gathered in Nairobi answers questions that datasets from richer countries cannot. It shows why data from under-studied places has its own value.
ImpactA community centre sets its own rules for its data
An urban Indigenous organisation explains how it governs data without legal power over it, as a steward and not an owner.
ImpactGiving data for good: where tax relief applies, and where it doesn't
Some data does the most good given away. In a few countries the gift can also bring a tax deduction, and in many it brings none.
ImpactWhen company data saved lives
Three times ordinary company data, made for something else, ended up doing public good.
ImpactSatellites find most of the world's industrial fishing fleet is not publicly tracked
Combining satellite images with vessel GPS records revealed activity at sea that neither source showed alone.
AI and training dataRobots collect mountains of data, then lose it in folders
An open-source tool aims to make robot sensor data findable and ready for training.
TechnologyInstagram says India reports its recommendations half as often
A taps-and-hides signal from users is now steering what a feed is allowed to show.
Data valueApple asks 200,000 people to grade a health chatbot
A study invites people to ask an AI their health questions, with a promise that the answers will not train a model.
Life sciencesHeart ultrasound images made from heart models
Researchers make moving heart scans from computer models, because real clinical sets are hard to build.
Policy and lawSame farms, five data sets, five different answers
Which data set a regulator picks can change which crops are told to cut their fertiliser.
Policy and lawItaly says your photo gallery needs its own yes
A regulator fined a social app for suggesting photos from phone galleries, saying one vague permission was not consent.
Life sciencesAuthors pull their own mouth-cancer data paper
A 37-author team withdrew a dataset paper after finding bugs in its script, saying the reported figures may mislead.
Policy and lawYou told a telehealth app it was private. Was it?
Five more US states join a federal case over health answers allegedly passed to advertising platforms.
MarketsA middleman can learn both prices without seeing either
A new paper shows a broker can learn the right buy and sell prices from yes-or-no answers alone.
Policy and lawA search giant must let publishers switch off its AI answers
The UK competition regulator has set a first rule: owners of web content get a say in how it feeds AI search.
AI and training dataEven coral experts disagree on about one label in twelve
A reef dataset built from 76 scattered sources shows what it takes to make labelled images fit for training.
Life sciencesBrain recordings from 7,272 people are rebuilt for training
A public set of brain-wave data keeps every owner's licence and still trains a model almost as well.
AI and training dataFourteen experts label bias in 448 Hindi news articles
A small open data set shows how a language left out of the tools gets its first measuring stick.
AI and training dataThe most accurate post moderator is often not the fairest
A new test set shows that judging posts well and judging them fairly are different skills.
Life sciencesCardiologists invent patients so heart notes can be shared
Doctors wrote artificial hospital reports in seven languages, so AI can learn from clinical notes with no real patient behind them.
Policy and lawSatellites can see the poachers, but the law says may
A study of marine park rules finds that data to catch illegal fishing is often useless in court because the wording is soft.
AI and training dataAI still struggles to write like one particular person
Six models were asked to mimic individual writers; the best one mostly copied.
Life sciencesReaders could not tell AI themes from clinicians' themes
Three AI models read patient life stories and found themes the researchers missed.
TechnologyTeaching an AI model to say it does not know
A new method nudges a few internal switches so a model declines to guess.
AI and training dataMachine-written tests catch more AI failures than human ones
A method that grounds test questions in real-world evidence makes chatbots fail more often on medical advice and self-harm.
Data valueNot every stale record is worth the cost of re-checking
A short paper shows that choosing which data to refresh by the task at hand beats refreshing blindly.
Life sciencesThin patient records learn to borrow from a map of medicine
A model decides, visit by visit, whether a patient record needs outside knowledge or can stand alone.
AI and training dataIgala gets open translation data that records dialect
A small sentence set for a Nigerian language shows why the mix of dialects in data changes results.
AI and training dataReweighting training data can shift harm to another group
A preprint finds a common fairness fix lowers missed cases for one group by raising them for another.
MarketsA broker offers to license a company's data to AI developers
SourceX of Los Angeles says businesses, including closed ones, can sell prepared data through one contract, without dealing with each AI lab.
AI and training dataA new corpus tests abuse detection in three languages
Detectors built on Northern data miss online abuse of women in Levantine Arabic, African French and Swahili.
MarketsA stock-photo firm's data licensing falls sharply in a year
Shutterstock's third-quarter results show how lumpy AI data revenue can be: one quarter a large deal lands, the next it does not.
Data valueImage checks cannot prove where a medical scan came from
A test of synthetic chest X-rays at dataset intake says provenance has to be recorded, not guessed from pixels.
Life sciencesPublished papers hold more assay data than the databases
An AI workflow lifts results out of papers and builds a larger, versioned database for protein degradation drugs.
ImpactA community centre sets its own rules for its data
An urban Indigenous organisation explains how it governs data without legal power over it, as a steward and not an owner.
Life sciencesA curated set and shared test make protein-model building comparable
CryoAtlas filters public structure records into one training set with a fixed benchmark.
AI and training dataA text-reuse test set built only from open sources
Researchers release a bilingual set for testing plagiarism detectors, with every source passage openly licensed or in the public domain.
Life sciencesA tool that cleans health records for AI is put to the test
A new benchmark checks a chat agent that turns raw patient records into datasets ready for AI, and reports where it falls short.
Life sciencesA public trial registry trains a simulator that ignores the drug
A new test shows models built on registry data can score well without using the one field that matters.
TechnologyFresh human data is what stops models forgetting
A new paper shows that models trained on model-made text lose rare knowledge first, and that a share of fresh data holds the line.
Sovereign and infrastructureIndia proposes that AI models travel to the data, not the other way round
A draft handbook from India's technology ministry sets out how data owners could let AI be trained without handing their data over.
Data valueThe data week so far: 1 to 3 November 2026
Nine data and AI stories from the first three days of November, each written from its source.
AI and training dataBig AI companies back one signal for a website to say no to training
One robots.txt token could replace separate rules for each company's crawler, though start dates differ and it covers training only.
AI and training dataSwitching off redundant parts of a language model cuts how much text it repeats
A preprint shows a way to reduce word-for-word recall of training text without retraining or changing the data.
TechnologyA wing near the ground gets its own simulation set
770 airflow simulations aim to test AI shortcuts for racing aerodynamics, though the download link is not yet live.
TechnologyOne aviation safety dataset built from 128 sources
A new paper merges national accident records into one deduplicated, traceable corpus.
AI and training dataA legal AI maker tests its agent without seeing client data
Legora's CTO explains how to improve an AI product when the data belongs to the customers' clients.
Life sciencesA health-records AI model holds up in hospitals it was never trained on
Researchers pretrained a model on records from four health systems and tested it on two others, measuring what screening would really look like.
Lost and foundWhen a startup closes, its code and conversations now have buyers
Firms that help startups shut down now sell what is left behind: the code, the documents and the working records. The prices are real, and so are the pitfalls.
Policy and lawEU governments agree their line on simpler data rules, with AI and research in view
The Council's position on the Digital Omnibus keeps the GDPR's definition of personal data, eases research and AI development, and folds Europe's data-sharing laws together.
AI and training dataEurope's privacy regulators tell AI scrapers to stay away from sites that say no
As consultation on their web-scraping guidelines closes, Europe's data protection authorities set out what AI developers must do before they collect from the open web.
Data valueA rare public price for health data: what one patient record is worth
A filing with the US markets regulator puts a figure on de-identified patient records, and it is rarely this plain.
Policy and lawA US appeals court says the market for AI training data is real
The first federal appeals ruling on AI training and fair use treats licensing data for training as a market an owner can lose.
Policy and lawUS regulator asks nine companies how personal data sets their prices and pay
The Federal Trade Commission wants to know whether what companies know about you changes what you pay, and what gig workers earn.
Data valueSold short: why so many owners part with their data for less than it is worth
The books say a company's data is worth nothing, and people price their own data by how the question is asked. Buyers know better.
TechnologyA model invents the storms short records never saw
Flood designers work from about ten years of rainfall. A new paper generates the extreme storms those records miss.
Physics and frontier scienceA detector's full record can tell more than the single number it is usually boiled down to
A new preprint finds that using the whole simulated record of a particle shower sharpens the energy reading by a factor of 1.7 to 1.9.
Life sciencesRegistry text alone carries some signal on which trials succeed
Language models read ten public fields per trial and beat the published benchmark, but not a trivial guess on every measure.
AI and training dataWhy old code is the hottest data in AI right now
AI teams want code that people wrote before AI started writing it. Many companies have years of it sitting in their repositories.
Data valueLease, license or sell: what each means for your data
Three ways to work with a buyer. The difference is who keeps a copy, for how long, and who owns it afterwards.
Lost and foundWhat happens to a company's data when it closes
The archive outlives the company. Code, drawings and records often sit untouched on servers long after the last day of trading.
Policy and lawTaking personal data out before a sale
AI buyers want what your data shows about the work, not about people. Here is what taking personal data out really means.
ImpactGiving data for good: where tax relief applies, and where it doesn't
Some data does the most good given away. In a few countries the gift can also bring a tax deduction, and in many it brings none.
Lost and foundLost in time: how company and research data quietly disappears
Nobody deletes it on purpose. It simply stops being looked after, and one day it is gone.
AI and training dataWhy AI still needs data made by people
AI teams are running short of good human-made data, and the open web is closing its doors. Licensed company data is one of the few new sources left.
ImpactWhen company data saved lives
Three times ordinary company data, made for something else, ended up doing public good.
Sovereign and infrastructureUK science advisers ask government to name its important datasets
Advice to ministers on AI sovereignty puts data beside computing and skills, and asks for access to be tested in health and creative data.
Sovereign and infrastructureChad validates a national plan that treats data as a sovereign asset
A workshop in N'Djamena has signed off a seven-part data governance strategy, with sharing, infrastructure and a data economy on the list.
MarketsA publisher now earns from licensing other publishers' work to AI
Wiley's annual report shows AI licensing growing, with a large part of it now coming from content it licenses on behalf of others.
MarketsAI licensing lifts an academic publisher's year
Bloomsbury's results show its academic and professional division growing on the back of licensing content for AI, a year after its first such deal.
Data valueInvestors tell accounting's standard setter: explain intangible assets before you book more of them
Research for the rewrite of IAS 38 finds users of accounts prefer better disclosure to putting more home-grown intangibles, data among them, on the balance sheet.
Data valueHidden in plain sight: the US starts counting what companies spend on their own data
A new estimate from the US national accounts treats a company's own data as an investment. The sums are large, and almost none of it shows on a company's balance sheet.
MarketsA licensed-data marketplace joins the plumbing of the web
Cloudflare buys Human Native, a UK marketplace that turns content into licensed training data, and folds it into the network that sits in front of much of the web.
AI and training dataPublishers agree a common language for licensing content to AI
Really Simple Licensing 1.0 lets a website state, in a form machines can read, how AI may use its work and on what terms.
Policy and lawEngland's High Court rules a trained image model is not a copy of its training pictures
Getty Images loses its remaining copyright claim against Stability AI in the first full UK trial over building an AI model.
Lost and foundThe Wayback Machine passes one trillion saved web pages
The Internet Archive marks a milestone in saving the web, and a reminder of how much of it would otherwise be gone.
Policy and lawEurope's Data Act takes effect, giving users a claim on the data their machines make
From today, people and businesses that use connected products in the EU can ask for the data those products generate.
AI and training dataEurope sets the form AI developers must use to describe their training data
The European Commission publishes the template general-purpose AI models must fill in under the AI Act, with its own place for licensed data.
Sovereign and infrastructureThe US calls high-quality data a national strategic asset
America's AI Action Plan asks for the world's largest AI-ready scientific data sets, and for federally funded researchers to disclose the data their AI uses.
MarketsA price tag for every page an AI crawler reads
Cloudflare lets website owners charge AI crawlers by the request, reviving an old web code for payment.
Policy and lawThe UK writes access to customer and business data into law
The Data (Use and Access) Act 2025 gives government the power to set up schemes that let customers and businesses get their data shared with services they choose.
Policy and lawThe US Copyright Office says licensing for AI training should be left to grow
Its report on generative AI training finds licensing deals spreading, and counts lost licence income as real harm to owners.
Data valueCanada finds companies make almost all the data they use, and buy very little
A Statistics Canada study counts data as the largest single kind of intangible investment in the Canadian business sector.
Data valueData joins the world's national accounts as an asset
The UN's statisticians back a new standard that counts data as capital, the first rewrite of the rulebook since 2008.
Lost and foundA law library saves a copy of every dataset on data.gov
Harvard's Library Innovation Lab releases an archive of the US government's public datasets, signed so each file can be proved genuine.
Policy and lawEurope's privacy regulators set the test for when an AI model is truly anonymous
The EDPB's opinion on AI models says training on personal data does not make a model anonymous by default, and spells out how legitimate interest can apply.
ImpactSatellites find most of the world's industrial fishing fleet is not publicly tracked
Combining satellite images with vessel GPS records revealed activity at sea that neither source showed alone.





































