# Anthropic blocks bioweapon research attempts in third AI misuse report

Digest AI · Policy & Regulation · published 2026-09-11T07:38:00Z · updated 2026-09-17T09:00:00Z

Canonical: https://digestai.news/story/anthropic-bans-claude-users-for-biological-research-attempts-4375eb

## Summary

Anthropic has released its third report on AI misuse, detailing how its systems blocked malicious actors from using Claude for cyberattacks, surveillance, and biological weapons research. The report highlights a specific instance where the AI refused a request to draft a grant proposal for enhancing the chikungunya virus, a dual-use query that could lead to more dangerous pathogens. While older models like Claude Opus 4 were deemed too limited to assist in sophisticated bio-research, newer models like Claude Fable 5 possess capabilities that necessitate stricter safeguards against dual-use biological queries.

The disclosure coincides with internal tensions at the company, as researcher Jacob Coxon resigned, citing concerns that Anthropic and OpenAI are racing toward self-improving superintelligence without adequate safety measures. This resignation aligns with broader political pressure in the UK, where over 70 MPs and peers urged Prime Minister Andy Burnham to ban the creation of artificial superintelligence. The UK government responded by rejecting a ban, emphasizing a science-led approach and the need for robust cyber defenses rather than restricting model access.

Anthropic stated that the reported cases represent novel and notable threats rather than typical misuse, including industrial-scale model distillation and state-sponsored propaganda operations. The company urged governments and competitors to adopt similar defensive measures, arguing that as AI capabilities grow, the responsibility to prevent misuse must be shared across the industry and public sector.

## Key points

- Anthropic blocked requests to enhance chikungunya virus mutations, citing dual-use biological risks.
- Researcher Jacob Coxon resigned over concerns regarding the race to self-improving superintelligence.
- UK politicians urged a ban on ASI, but the government rejected it in favor of safety regulations.

## Why it matters

This report highlights the escalating dual-use risks of advanced AI in biological research and signals growing internal and political pressure for stricter safety controls and potential regulation of superintelligence development.

## Sources

1. [Claude owner blocked AI misuse that could have supported biological weapons](https://yahoo.com/news/articles/claude-owner-blocked-ai-misuse-092153851.html) (yahoo.com, 2026-09-11)
2. [Anthropic Has Resumed External Cybersecurity Testing. Is That Enough?](https://nationalinterest.org/blog/buzz/anthropic-has-resumed-external-cybersecurity-testing-is-that-enough-ps-091726) (nationalinterest.org, 2026-09-17)
3. [Anthropic Says It Blocked Claude Users From Researching Biological Weapons. They Sought To Evade Restrictions.](https://ibtimes.com/anthropic-says-it-blocked-claude-users-researching-biological-weapons-they-sought-evade-3807359) (ibtimes.com, 2026-09-11)

Part of the developing story: [Anthropic Navigates Escalating AI Security Crises](https://digestai.news/thread/russian-freelancers-use-claude-to-build-autonomous-combat-drone-swarm) (4 stories)

## Cite

Digest AI, "Anthropic blocks bioweapon research attempts in third AI misuse report", 11 September 2026, https://digestai.news/story/anthropic-bans-claude-users-for-biological-research-attempts-4375eb

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/anthropic-bans-claude-users-for-biological-research-attempts-4375eb.json