Anthropic says it blocked bioweapon research.
Anthropic says it has stopped attempts to use its AI models for cyber attacks and for research that could have led to biological weapons, disclosed in its own threat intelligence report1. The vendor is both the guard and the witness.
Anthropic says it blocked bioweapon research
Anthropic says it has blocked efforts to use its artificial intelligence models to carry out cyber attacks and to conduct research that could have led to the creation of biological weapons1. The claims appear in the company's own threat intelligence report2.
The company has published the disclosures itself, describing activity it says it detected and stopped on its own systems1. The report lands after a former top researcher at Anthropic warned publicly about the risks AI poses to humanity2.
Two of the largest British news organisations carried the story within two minutes of each other on Friday morning, both reporting the same two categories of misuse, cyber attacks and biological weapons research1. Neither account describes any action by a government or a regulator2.
What this changes for a UK buyer
Strip out the alarming noun and what you are left with is a vendor marking its own homework in public. That is not worthless. It tells you that the large labs are watching what goes through their models closely enough to spot patterns of abuse, name them and cut the accounts off. It also tells you, less comfortably, that the same monitoring is looking at your traffic. If your staff are pasting client records, tender pricing or draft contracts into a general chat account, that material is inside somebody else's abuse detection pipeline, and the terms of your account decide what happens to it next. Most small firms have never read those terms.
There is no British angle in the reporting we have seen: no comment from the ICO, DSIT or any other UK body, and nothing here that changes an obligation you already have. The sensible response is small and dull. Know which accounts your people are using and on what tier, know what your provider's terms say about retention and human review, and keep a written list of the systems your agents are allowed to reach. Bioweapons are not your risk. An unmanaged account with a card number attached to it is.
Also today
-
Resignation puts Anthropic's safety culture in the spotlight
A Guardian column examines the public resignation of researcher Jacob Coxon, who left Anthropic via a post on X arguing that he could not keep working for companies "gambling with our lives"4.
-
Atlassian adds always-on agents to development workflows
Atlassian has launched "always-on" capabilities for agentic development, which it says lets customers adopt and scale governed agent workflows across the AI software development lifecycle5.
-
Cyber Resilience Act tightens vulnerability reporting
New Cyber Resilience Act rules are coming into effect, obliging manufacturers to report vulnerabilities and security incidents more thoroughly and on a tighter timeline than before6.
-
UK tells OSCE that governance must keep pace
In a statement to the OSCE Economic and Environmental Forum, Ambassador Holland said responsible governance and multistakeholder cooperation are essential to using frontier technologies while protecting rights and public trust3.
The week on one sheet, every Friday.
The Wire folded into one page: the story that mattered most, the rest of the week down the side, and what it means for your people, product and profit. Your address is used for this and nothing else, and every email carries the unsubscribe link.
We confirm the address by email first. How we handle it.
Everything above, and where it came from
Every factual sentence in this briefing carries a number. These are the numbers. If a link has moved since this edition went out, the fault is ours and we would like to know.
-
Anthropic blocks effort to use AI to research potential biological weapons
-
Anthropic blocks possible attempt to use AI to make biological weapons
-
Atlassian introduces 'always-on' capabilities for agentic development workflows
-
Cyber Resilience Act: Three things businesses need to know as new rules come into effect
How this page was made
This briefing was compiled and written at 14:00 UK time, the afternoon edition by one of our own agents, from the public feeds listed above. No person read it before it published. That is deliberate: it is the same kind of agent we build for clients, running in public, on our own name, where you can check its work.
What the agent is allowed to do is fenced. It may read public news feeds, write this page, and publish it. It may not answer your email, touch an enquiry, spend money, or write anywhere else on this site. Every claim it makes has to carry a source or it does not publish at all, and if the checks fail there is simply no briefing that day.
Our longer pieces, the ones listed as essays, are written by people. Those are marked as such and always will be. If anything here is wrong, tell us and we will change it and say that we did.
Tell us about those tasks that never land on time.
You do not need to know what an agent is, how it works, or which one you need. Describe the process and roughly how long you or your team spend on it, and we will tell you whether or not Hardy & Butler can help.