Who Actually Owns the Books That Trained Your AI?
AI companies are sitting on a legal question they’ve quietly hoped courts would ignore. This week, that question is getting harder to avoid: did anyone have the right to feed millions of published books into AI systems without asking the authors first? Meanwhile, new tools are quietly reshaping how we manage our homes and our privacy — and not everyone is happy about it.
AI Companies Trained on Your Favorite Books. Authors Want Answers.
Copyright law — the set of rules that decides who controls how creative work gets used — has a long history of struggling to keep up with new technology. It happened with the photocopier, the VCR, and the internet. Now AI is next in line, and the stakes feel unusually high for anyone who has ever written anything for a living.
Here’s the situation: AI developers built their large language models, which are systems trained on enormous amounts of text to learn how language works, by feeding them vast libraries of books. Many of those books were copyrighted, meaning the authors never signed away the right to have their work used this way. Nobody asked. Multiple lawsuits are now working their way through courts, with authors arguing this amounts to theft and companies arguing it qualifies as “fair use” — a legal principle that allows limited use of copyrighted material without permission, typically for purposes like education or commentary.
Think of it like this. If you photocopied a novel and sold the copies, that’s clearly illegal. If you read a thousand novels and then wrote your own, that’s clearly fine. AI training sits in a murky middle: the model doesn’t reproduce the books directly, but it learns from them in ways that can directly compete with the authors who wrote them. Courts have never had to draw that line before.
For everyday readers and writers, TechCrunch’s breakdown of the ongoing legal battles makes clear that the outcomes here could fundamentally change what AI companies are allowed to build — and how much authors might eventually be owed.
Why this matters: If courts rule against AI companies, the cost of building these systems could rise sharply. If they rule in favor, it may become much harder for writers to protect their work.
“Copyright law hasn’t caught up with how AI companies use published works.”
This Calendar App Wants to Run Your Entire Household
Scheduling apps usually do one thing: remind you where you need to be. Linkdaze is reportedly trying to do something more ambitious — manage the rhythms of an entire home.
According to TechCrunch, Linkdaze has launched a calendar app that goes beyond appointments. It reportedly layers in household management features like meal planning, and bakes in AI capabilities without charging extra for them. That last part matters more than it might seem. Most apps have started treating AI features as premium add-ons, so a product that bundles them into the base experience is swimming against the current.
The idea is straightforward: rather than juggling three separate apps for your schedule, your grocery list, and your family’s weekly routines, one tool holds it all together. The AI component reportedly helps by suggesting and coordinating across those categories automatically, rather than waiting for you to enter everything by hand.
For busy households, the appeal is obvious. The question is whether the execution lives up to the concept. TechCrunch covered the launch, though other major outlets have not yet independently confirmed the details, so some caution is warranted before reading too much into the claims.
Why this matters: The real competition here isn’t other calendar apps — it’s the friction of modern family life, and AI that can reduce that friction without a subscription tax is genuinely interesting.
A Surveillance Company Is Asking for Trust. Critics Aren’t Ready to Give It.
Flock Safety makes technology that tracks vehicles using license plate readers — cameras that automatically log which cars pass by and when. The company has sold these systems widely to police departments and neighborhoods. Now, according to TechCrunch, the backlash is growing loud enough that the CEO is publicly asking for “compromise.”
What that compromise looks like in practice isn’t entirely clear. The concern from critics isn’t just theoretical: a network of cameras that logs vehicle movements can, over time, build a detailed picture of where someone goes, who they visit, and when they sleep at home. That data exists whether or not anything criminal ever happens.
The CEO’s call for dialogue is a response to that discomfort, but critics argue that the problem isn’t how the tool is used on any given day — it’s how much it could reveal if the rules change. TechCrunch’s reporting frames this as a values dispute that no amount of corporate goodwill can easily resolve.
Why this matters: Surveillance tools rarely get rolled back once they’re installed. The conversation happening now is likely the last easy moment to shape how they’re governed.
Also Happening in AI
Researchers published VIALS, a new benchmark — essentially a standardized test for AI systems — focused on interpreting real scientific images from life sciences research, a surprisingly difficult task for current models. On the privacy front, OpenAI announced it’s offering API customers the option to use its services with zero data retention, meaning the company won’t store conversations at all. Separately, a developer on Reddit demonstrated a working watermarking technique for AI-generated text, which subtly embeds invisible patterns to help identify machine-written content. And a practical tutorial from Towards Data Science argued that AI tools analyzing legal case files work better when documents are organized relationally rather than processed as isolated PDFs.
What to Watch
The copyright lawsuits against AI companies are approaching critical phases in several U.S. courts, and a ruling in any one of them could set a precedent that reshapes the entire industry almost overnight. Watch specifically for how judges define “transformative use” — whether training an AI on a book counts as doing something new with it, or simply copying it at scale. The answer will matter far beyond the courtroom.