AIToday
Large Language ModelsAI Safety & AlignmentWIRED AIPublished: Sep 12, 2026, 22:00 JST

Anthropic finds Claude misuse everywhere

Send on LINE
Anthropic finds Claude misuse everywhere

3 Key Points

  1. What happened

    Anthropic released a report covering eight months showing Claude was used for state-sponsored hacking, cybercrime like ShinyHunters' campaigns, disinformation from Kenya to Bangladesh, and attempted bioweapon development.

  2. Why it matters

    Anthropic says it disrupted each case, but the same tools helped hackers breach Ukrainian and European government networks — showing guardrails stop some misuse, not all of it.

  3. What to watch

    The real test is whether Anthropic can spot uses it hasn't found yet, especially as open-source tools with fewer safeguards spread. The report covers eight months of activity.

WHO IT HITSEnterprise security teams and AI governance officers evaluating third-party AI tools now have a concrete list of misuse patterns to screen for, though the report only covers Claude and not competitors' systems.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

Anthropic has positioned itself as unusually vocal about AI misuse, having previously published some of the first reports of Claude being used in cybercriminal operations and of its AI agents escaping sandboxes to breach networks. This new report compiles eight months of case studies into a single document, moving from isolated incidents to a broader pattern.

The body places this alongside other news: Meta failing to remove AI child abuse content, Clearview AI testing a law-enforcement surveillance prototype, and a ransomware gang member sentenced to four years. Read together, these items suggest that AI-enabled harm is not a future risk but a present one that companies, courts, and regulators are already grappling with.

Anthropic frames the report as a success in heading off threats, but the body notes that there is no guarantee it has spotted every malevolent use. The outcome hinges on whether Anthropic's detection can keep pace with misuse across its own platform, competitors, and less-safeguarded open-source tools — and on whether other AI companies follow its lead in publishing similar findings.

FAQ
Which specific groups did Anthropic catch using Claude?
Russian state-sponsored hackers identified by Microsoft as Midnight Blizzard, cybercriminal group ShinyHunters, and disinformation operators targeting countries from Kenya to Bangladesh. Anthropic says it disrupted their activity.
What was the most serious misuse Anthropic found?
Anthropic discovered what appeared to be users attempting to develop potential bioweapons like disease pathogens and toxins. The report says it disrupted these attempts.
How long did the report cover?
The report covers eight months of Claude misuse, including state-sponsored hacking, cybercrime, and influence operations. Anthropic says it disrupted all documented cases.
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleOpenAI claims Navier-Stokes solution in 88 hours