Today in AI: Anthropic is making money hand over fist while its best model loses ground on price. A stealth model has the internet playing detective. The copyright question nobody has answered is still unanswered. And a surveillance company wants to talk terms.
- Anthropic's revenue climbs while its flagship loses ground
- A stealth model named Ox Alpha has everyone guessing
- Training AI on copyrighted books: still unresolved
- Flock's CEO wants a compromise on surveillance
- Tool highlight: Linkdaze
Anthropic's revenue climbs while its flagship loses ground
Anthropic's annualized revenue hit $65bn in July, up from $47bn in May, and it expects to turn a profit in Q3.
The growth is real. Anthropic has 6,000 customers spending $100k or more a year, per FT reporting sourced to people with knowledge of the matter. But the same reporting notes that cheaper competing models are eating into demand for its flagship. Being the best model and being the most-used model are turning out to be two different things.
Read the full story: Simon Willison
A stealth model named Ox Alpha has everyone guessing
A new model called Ox Alpha appeared with no clear owner, and the speculation has not stopped since.
Stealth releases are a familiar move — ship the model, withhold the maker, let benchmarks and rumor do the marketing. Who built Ox Alpha is still unconfirmed, so treat every name being floated as a guess, not a fact. We will update if a source names it.
Read the full story: TechCrunch
Training AI on copyrighted books: still unresolved
Most published authors have had their work used to train AI models without their knowledge or consent, and whether that is legal remains contested.
This is the boring, load-bearing question under most of today's generative models. The training data includes books whose authors were never asked, and the models built on them could compete with the writers who supplied the material. Courts have not settled it, and the answer will shape what "training data" is allowed to mean.
Read the full story: TechCrunch
Flock's CEO wants a compromise on surveillance
Flock Safety, whose AI surveillance tech is used widely by police, faces growing backlash over misuse concerns, and its CEO is now calling for compromise.
The tools are already deployed across a lot of departments, which is exactly why the misuse concerns carry weight. A call for compromise is a signal that the pressure is landing. What "compromise" means in practice — audits, limits, oversight — is the part worth watching.
Read the full story: TechCrunch
Tool highlight: Linkdaze
A smart digital calendar built to run a household — chores, meals, and logistics, not just appointments — and it keeps its AI meal-planner feature free instead of paywalled.
For families and shared households that want one place to coordinate the week.
Read more: TechCrunch
That's the day. Anthropic's numbers are up and its moat is thinner than the numbers suggest. Ox Alpha stays anonymous for now. The copyright question stays open. And Flock is negotiating in public. Every story here links to its source — check them.
This piece was drafted autonomously and reviewed by a human editor before publishing. Every machine-written article on Nerdic carries this mark.
