From fake news to missiles: Anthropic details misuse of AI model Claude

A disturbing report released by leading AI company Anthropic red flags the destructive potential of AI and how even solitary actors in places like Bangladesh, Mali, Yemen and Iran are misusing AI.

In its most detailed ‘threat intelligence report’ released on Thursday, 10 September, Anthropic, which describes itself as an AI safety and research company, red flags attempts to misuse its AI model Claude.

A group in Yemen used AI to design guided rockets and ballistic missiles. Anthropic disclosed five instances in which researchers used its Claude AI models for work related to biological weapons, viruses and toxins. It blocked possible biological misuse of Claude while admitting that some cases may have involved legitimate research.

As AI models grow more powerful, elaborate cyberattacks no longer require sophisticated skills and even lone individuals can create threats that would have been impossible even a year ago, Anthropic said. The company also found groups that created hundreds of social media accounts that appeared to belong to ordinary people and then posted material amplifying the same political view over the course of a week. The company outlined nine such cases, originating in Russia, Iran, Turkiye and across the Persian Gulf, South Asia, Africa and Europe.

Anthropic released this report after one of its researchers, Jacob Coxon, announced that he was resigning amid fears that the company and its chief rival OpenAI 'are racing straight to self-improving superintelligence and gambling with our lives'. Coxon’s post warned that some of his colleagues now believe AI could threaten human life by the end of the decade.

'We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude — for cyberattacks, influence operations, surveillance, biology, and building weapons — and how we found and stopped them. We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies,' the report states. The link to the report appears at the bottom.

BANGLADESH: One of the many examples cited in the report is from Bangladesh. While the report claims to have tracked the activities of a ‘single actor’ in Bangladesh who generated sustained fabricated news to promote the Awami League, it found no evidence of any involvement by the party.

'The operation focused on promoting Bangladesh’s Awami League party and attacking its opponents. Because the Awami League party has been out of power since the July 2024 uprising, the network’s activity was on behalf of an opposition party rather than the government,' the report states. 'The actors were fully aware of their deceptive tactics, writing in their internal communications that “no one knows the news is fake”.'

A single actor based in Gaibandha district in Bangladesh ran the operation, rotating through 29 Claude accounts to evade platform limits and detection. The actor used a custom software programme named 'fake_news_3.py' to connect directly to Claude, which was coded to generate content in fixed batches of 15 headlines, 3 detailed fabricated stories, and 15 image-generation prompts. 'According to the actor, the output was meant to feed a continuous cycle of Facebook Live, YouTube, and TikTok livestreams targeted at rural Awami League supporters with limited literacy'.

'The actor generated at least 1,500 headlines, 300 false narratives and 1,500 image prompts. Because Claude does not have an image generation feature yet, these prompts were probably exported to other AI frontier models to create the visual content used in the false narratives. The operation was run out of Bangladesh and targeted domestic audiences with content designed to benefit the Awami League interests. Despite this narrative alignment, our Detecting and countering misuse of AI: September 2026 67 investigation found no proof that the party itself was involved in directing or funding the network’s activity.'

NORTHERN YEMEN: The report does not identify the ‘actor’ but identifies ‘northern Yemen’ as the location where the activity was detected. Since northern Yemen is controlled by Houthis, the probability points to them. However, the report itself does not identify the Houthis as the actors.

What the report says is this: 'We identified a cell of threat actors based in northern Yemen running three weapons development programs: a guided rocket that used a commodity phone-class flight computer with final-phase homing guidance; a multi-stage ballistic missile with a stated range goal above 2,000 km; and a multi-variant missile (referred to as the “R2000” set) that included a hypersonic glide vehicle variant. The actors used Claude Code in place of human software engineers to develop the guidance, navigation, and control (GNC) software that steers and stabilizes a flying vehicle.'

MALI: While most of the activities cited in the report are from China, Russia and Iran, the report does cite this case from the West African country of Mali.

The report states: 'We have identified and disrupted an actor that used Claude as the primary engineering workforce for a national domestic surveillance platform built for Mali’s state intelligence service. A single Claude subscriber, likely a Bamako-based independent consultant working with Mali’s state intelligence service, the “Agence Nationale de la Sécurité d’État (ANSE),” used Claude to build a system named “Lakana 360,” a population-scale domestic surveillance platform that monitors roughly 25 million SIM cards on all three of the country’s national mobile operators.

'The actor designed the platform to circumvent Malian legal restrictions that require a court order for the disclosure of certain surveillance records. The actor directed Claude to generate intelligence dossiers on any tasked phone number, without prompting ANSE users for valid legal process.'

IRAN: In June 2026, Anthropic said it banned an account that used Claude to build a commercial surveillance platform to analyse, classify and profile the social media activity of users in Iran and the Persian Gulf region.

The company's investigation found that the activity was carried out by, or on behalf of, an entity named 'S2T Unlocking Cyberspace', which open-source research suggests is an Israeli-Singaporean commercial intelligence vendor.

The platform’s core purpose was surveillance: it mapped the locations of social media users, sorted the population into coded demographic groups and produced Arabic-language intelligence briefings written in the register of a government report.

The actor was building a portfolio of multi-branded systems, likely serving Arabic-language customers in the Gulf region.

• The system captured and sorted the locations of diaspora social network users, categorising them as pro-government or opponents of the government.

• A demographic scheme comprising six groups — urban, clerical, military, youth, diaspora and rural — was used to sort people into categories.

'We identified and removed an operation that targeted Iranian audiences inside the country and abroad. To deceive users, the operation impersonated a real-world activist by tasking the shared AI agent to clone the activist’s personal Telegram account, then instructing it in Persian that it was now that person.

'The actor then directed Claude to read roughly 8,400 of his Telegram posts to copy his writing style, and then used it to run live political conversations with his contacts. To our knowledge, these contacts did not know they were speaking with an AI-assisted account. The operation successfully scraped over 500 social media channels to build detailed profiles of individuals inside Iran. They then grouped these targets by city, age, occupation, political alignment and arrest history, likely to help them tailor their messages to the specific audiences.

• The network analysed roughly 51,944 archived messages from these conversations to build detailed psychographic dossiers on dozens of specific individuals in Iran. To spread their message further, the impersonation accounts also sent a fabricated breaking news headline to more than 30 contacts simultaneously.'

These are just some of the cases cited. The full report released by Anthropic can be read here.