Anthropic: Claude was misused in paid campaign targeting Malaysian voters
Anthropic, the US company behind the Claude AI chatbot, says it has shut down an account linked to a paid influence operation that primarily targeted users in Malaysia. The case was detailed in the company’s latest threat intelligence report, which looks at how its AI tools have been misused for cybercrime and influence operations.
According to Anthropic, the operation was run as a commercial service that allowed paying clients to carry out political influence campaigns. While the platform presented itself as a defensive cyber intelligence and counter-disinformation service, Anthropic said its investigation found links to BBS Bilisim Teknolojileri, a technology company based in Istanbul, Turkey.
Documents examined by Anthropic described the platform as a “military-grade, AI-driven, real-time political operations ecosystem”, with tools designed to create and spread political content at scale.
The operation involved three main components:
- A network of roughly 1,000 fake accounts on X, formerly known as Twitter
- A fake news website called “Malaysia Pulse”
- Fabricated intelligence dossiers containing false allegations
The system also used real census and electoral data, along with millions of voter records, to build profiles covering all 222 parliamentary constituencies in Malaysia. According to Anthropic, the operators used the information to identify politically sensitive issues, particularly those involving race, religion and royalty.
The fake social media accounts were designed to appear more like genuine users. Anthropic said accounts would first go through a “warm-up” period before being used for campaigns, while measures such as changing cookies and IP addresses were used to reduce the chances of detection.
The platform also allowed users to specify the amount of artificial engagement they wanted. In one example seen by Anthropic, a user requested one million artificial views for content supporting Malaysia’s sitting Prime Minister. However, Anthropic did not identify who made the request or find evidence linking the operation to the Prime Minister or his office.
At the same time, fabricated intelligence dossiers were used to spread false allegations against an opposition politician and civil-society organisations. Anthropic said the allegations were made up by the operators, while its own research found no corroboration for claims made against named individuals.
The “Malaysia Pulse” website was another part of the operation. According to Anthropic, the site took reports from legitimate Malaysian news outlets, rewrote them using AI and published them under fabricated journalist names. It also republished material from Russian and Chinese state-aligned media while removing the original attribution, making the articles appear to come from an independent local source.
Despite the scale of the setup, Anthropic found limited evidence that the campaign managed to reach a genuine audience. The company rated it as Category Two on its Breakout Scale, meaning the content appeared across multiple platforms but showed no evidence of breaking through into authentic communities or gaining meaningful organic traction.
Read more of our articles below!
