KUALA LUMPUR (Sept 11): US-based artificial intelligence (AI) firm Anthropic said it has disrupted the operations of a commercial election-manipulation platform that used its Claude AI models to target voters across Malaysia.

In a threat intelligence report published on Sept 10, the California-based AI research and safety company revealed that the network consisted of roughly a thousand fake X social media accounts, a fake news outlet, and a series of fabricated dossiers.

The platform "posed defensive cyber intelligence and counter-disinformation tooling outlet", which Anthropic traced to an Istanbul-based technology company named BBS Bilisim Teknolojileri, which sold access to the platform as a paid "influence-as-a-service" tool.  

"According to the threat actor's own documentation, the infrastructure was marketed as 'military grade, AI-driven, real-time political operations ecosystem'," the report noted.

How the election-manipulation platform used Claude:

Source: Anthropic report: Detecting and countering misuse of AI: September 2026

It did not specify exactly when the Malaysia-focused operation took place or how long it ran. The report covers disruptions conducted by Anthropic on cyber operations, surveillance, scams and other forms of misuse of its models between December 2025 and August 2026.  

According to Anthropic, the actor used Claude models to build an AI-powered voter profiling platform that ingested census and electoral data across all 222 parliamentary constituencies in Malaysia.

The system was engineered to target sensitive social and political fault lines, including race, religion and royalty.

The network controlled a “fabricated media outlet” and “inauthentic YouTube channel” called 'Malaysia Pulse' that used AI to scrape and rewrite legitimate local news articles under fake bylines.

It also republished articles from Russian and Chinese state-aligned foreign outlets, including TV BRICS, Xinhua, Sputnik/RIA and CGTN by stripping out the state attribution to present them as “independent Malaysian reporting”.

The Malaysian operation is part of a broader global trend of how Claude was used for malicious activity in influence operations, including for Russian propaganda operations in Moldova and Africa, demographic profiling of Iranians and Chinese, and the amplification of AI-generated articles across 70 fake outlets in 20 languages using social media bots. 

The report also presented case studies in harm areas such as surveillance, scams and fraud, biological misuse, conventional weapons development, and illicit distillation — industrial-scale, covert campaigns to extract a model's capabilities and replicate them elsewhere without authorisation. 

The platform's 1,000 automated X accounts were equipped with "warm-up" mechanisms — such as regularly renewing cookies and IP addresses — to mimic authentic users and evade platform detection systems.

The dashboard allowed users to tune how many artificial views a target should receive, with Anthropic observing one request "in support of the sitting Malaysian prime minister" for one million artificial views on his account.

Anthropic noted that engagement figures were self-reported by the actor and that it found no evidence connecting the operation to the prime minister or his office.

The platform was also used to draft fake intelligence files containing false allegations against a Malaysian opposition politician and local civil society organisations.

Where Claude refused requests that it identified as political defamation, the actor negotiated sanitised wording to bypass safety controls. The operator also pursued a contract with Malaysia's national communications regulator, according to Anthropic, though it found no evidence that the effort succeeded.

All accounts associated with the group have been terminated, Anthropic said, as it upgraded its safety filters to catch similar prompt negotiation attempts and shared threat intelligence with authorities and industry partners.

The Malaysian Communications and Multimedia Commission (MCMC) has issued a statement saying it takes the matters raised in Anthropic's report seriously, while denying any involvement in the activities described in the report. 

"For clarity, activities of the nature described in the report do not form part of MCMC’s functions or mandate, and MCMC does not engage in such activities or arrangements. 

"MCMC is examining the report and will seek further information and clarification from the relevant parties to establish the facts and assess the matters raised. Based on the information and evidence available, MCMC will determine the next appropriate course of action within its remit, in coordination with the relevant authorities where necessary," it said.

Read also:
MCMC reviewing Anthropic report on alleged election-manipulation platform, denies any involvement