PUTRAJAYA: Malaysia was targeted in an artificial intelligence (AI)-driven commercial political manipulation operation involving about 1,000 fake X accounts, voter data and constituency-level targeting, according to a report by AI company Anthropic.

Anthropic, the company behind the Claude AI platform, said it identified and removed an account that had used Claude to operate a commercial election-manipulation platform targeting Malaysian users.

In its report, Detecting and Countering Misuse of AI: September 2026, Anthropic said the platform was linked to Istanbul-based technology company BBS Bilisim Teknolojileri, which allegedly sold access to an “influence-as-a-service” operation.

The platform was marketed by its operators as a “military-grade, AI-driven, real-time political operations ecosystem” and was reportedly built using Claude.

According to Anthropic, the operation comprised roughly 1,000 fake X accounts, a fabricated news outlet called “Malaysia Pulse” and a series of fake intelligence dossiers.

The operation allegedly used census and electoral data to profile and target voters across Malaysia’s 222 parliamentary constituencies, focusing on sensitive political and social issues including race, religion and the institution of the Malay rulers.

Anthropic said the network’s fake accounts were designed to artificially boost engagement and evade social media detection systems.

The operators also allegedly used an AI rewriting system to generate content for the fabricated news outlet and produced false allegations targeting an opposition politician and civil society organisations.

Anthropic said the allegations contained in the dossiers were entirely fabricated.

The company classified the campaign as Category Two on its Breakout Scale, indicating that the operation’s assets were distributed across multiple platforms but there was no evidence that the campaign had broken into authentic online communities.

The report also said the operators used Claude Code to develop custom dashboards for managing and monitoring the network of fake accounts.

The dashboards tracked metrics including likes and views generated by the accounts for specific targets. In one case involving the social media account of a senior Malaysian government official, the system recorded figures in the millions.

Anthropic stressed that such figures were self-reported by the operators through their own tools and could not be independently verified.

The company also alleged that the platform could be configured to generate a specified number of artificial views, including a request to produce one million fake views for a target.

Anthropic said it detected and removed the account involved after conducting an internal investigation. It subsequently used the indicators identified during the investigation to map the wider operational network and disrupt similar attempts to misuse its AI systems.

The findings highlight growing concerns over the use of generative AI to automate political influence operations, fabricate information and manipulate online engagement at scale.