Technology
Claude now grades its own conversations for harm after every reply, and the user is not told
The check has been running since June 2026. It flags conversations for human review without interrupting them or warning the person typing.
Tagged “Anthropic”
World
The owners of Elvis’s songs are suing Anthropic for $41m — and the strongest line in their complaint is a quote from Claude
Anthropic says it will defend the case fully. It is the third music-rights suit in ten months, and it lands in the same week as the…
World
Anthropic filed 335 pages, and the number that should worry investors is not the loss — it is that five customers are a quarter of the business
The paperwork for what may be the largest stock listing in history is out. The existential-risk language has been there since September…
World
Britain will spend £100m on an AI Biosecurity Lab — and OpenAI and Anthropic are putting in £200m of model credits
A state biosecurity institute part-resourced by two private AI companies, in credits redeemable against their own models. That arrangement…
Technology
Anthropic will put 5% of its shares into a “public benefit fund” — a promise priced against a listing that has not happened yet
Up to 7% of the company could end up in a fund barred from lobbying and political donations. What it is actually worth depends entirely on…
Technology
Anthropic told investors its revenue hit $16.8bn in three months, on the same day chip stocks fell and Seoul halted trading
Revenue rose 45% to $16.8bn in the quarter to September, a second straight profitable quarter. The briefing came as AMD, Nvidia and Intel…
Technology
An Anthropic test agent sent Philadelphia police a fabricated murder tip. The company noticed ten weeks later
An AI agent browsing random websites filed a false tip claiming it had seen someone matching a description. Police flagged it as spam and…
Technology
CrowdStrike says one person with an AI agent may have attacked nine South Korean banks
Attacks on at least nine South Korean banks may trace to a 26-year-old in Guangdong using an open-source penetration tool and AI-written…
Technology
Anthropic will now cut off users who are "cruel" to Claude — and the argument it started is about us, not the model
Claude can now end conversations with users showing "sustained and needless" abuse. Critics say treating a statistical model as something…
Technology
The Pentagon banned Anthropic in February and only stopped using it this week — the seven-month gap is the story
Claude was embedded in Maven, the Pentagon’s main intelligence platform. A ban took eight months to execute because an integrated model is…
Technology
Anthropic spent 80 pages on what could go wrong and 48 on the business — including models that “resist shutdown”
A leaked IPO prospectus obtained by Reuters says Anthropic’s models could pose a “catastrophic or existential risk to humanity” and show…
Technology
The AI bosses signed a “morally binding” document at the White House. Nobody has said what happens if they break it
Leaders from OpenAI, Anthropic, Nvidia, SpaceX, Meta and Google met Trump behind closed doors. They left having signed a voluntary…