Anthropic, the company behind the artificial intelligence (AI) assistant Claude, said on September 10 that it had disrupted an operation that used AI to produce fabricated Bengali news stories promoting the Awami League and attacking its political opponents.
The operation was run by a single operator in Gaibandha district who used 29 Claude accounts over about 16 months to evade restrictions and detection, according to an Anthropic report published on its website titled, "Detecting and countering misuse of AI: September 2026".
Anthropic said the operation generated at least 1,500 headlines, 300 false narratives and 1,500 prompts for images. The material was intended for distribution through social media and video platforms.
The company said it found no evidence that the Awami League directed or financed the operation. It also said some of the narratives were consistent with pro-Indian geopolitical interests, but investigators found no evidence that the campaign was directed or funded by any government.
According to Anthropic, the operator knowingly presented fabricated material as genuine news. In communications examined by the company, the operator wrote that people would not know that the news was fake.
The operation was designed to produce a continuous stream of content for Facebook Live, YouTube and TikTok. The instructions called for material that was “hot and aggressive” and written in simple language that would be easily understood by rural audiences, Anthropic said.
“The operation’s content was uniformly pro-Awami League and targeted the Bangladesh Nationalist Party (BNP), Jamaat-e-Islami, the National Citizens Committee, the interim government, and student protest leaders. The network used fabricated smear campaigns to accuse these opponents of being foreign agents and advancing Taliban-style governance,” Anthropic stated.
Anthropic said the operation relied heavily on automation rather than manual production. A programme called “fake_news_3.py” was configured to connect directly to Claude and request a fixed batch of material. Each request produced 15 Bengali headlines, three detailed fabricated stories and 15 English-language prompts for accompanying images.
The material was then passed through cloud-storage folders and converted into audio and video, according to Anthropic. A separate programme automated bulk uploads to YouTube and scheduled their publication. The operator also used a third-party software automation service that concealed the operator’s internet address.
Image
A screenshot of a Google Drive folder that contained videos produced by the operation before they were distributed. Image: Anthropic.
The company said the image prompts were probably sent to other AI models because Claude did not generate images itself. It did not identify which other AI services may have been used.
Anthropic said it banned the Claude accounts associated with the activity after an internal investigation. It also shared technical information with distribution platforms and other partners and introduced additional detection measures intended to prevent the operator from returning with new accounts.
Anthropic’s report does not establish how widely the fabricated material circulated, how many people viewed it or whether it influenced political behaviour.
