x.com/VoiceofSharif/status/2098363309607264323Anthropic’s Threat Intelligence Report exposes how both the Islamic regime and the MEK terrorist group independently misused the Claude AI model between December 2025 and August 2026.
🔴
Pro-IRGC PropagandaAnthropic shut down three networks run by organizations of the Islamic regime in Iran.
The Strategy: They used Claude to translate state intelligence materials and automate propaganda plans.
Media Laundering: They fed MEK-affiliated articles into Claude and ordered the AI to rewrite them. This disguised the text as the voices of ordinary, independent citizens online while subtly pushing pro-IRGC narratives.
🔴
Group GTG-84006 (MEK Terrorist Campaign)In a completely separate discovery, Anthropic dismantled a highly coordinated network run by the MEK terrorist group that used Claude for recruitment and political attacks.
Identity Cloning: The group fed 8,400 real Telegram posts from an actual activist into Claude, forcing the AI to clone his exact writing style and voice.
Fake Live Chats: They deployed this cloned AI persona to hold live political chats with the activist's real contacts on Telegram to recruit them without their knowledge.
🔴
TARGETING PAHLAVIThe MEK network specifically used Claude to launch aggressive political attacks against the Pahlavi camp and monarchist groups. They ordered the AI to use the slogan "Neither Shah Nor Sheikh" to attack both the current regime and the pre-revolutionary monarchy simultaneously. They even created and circulated an AI-altered video designed to attack a member of the Pahlavi family to systematically degrade their public standing.
Resource Theft: To fund this malicious operation, the MEK stole AI API keys from legitimate businesses to hijack their computational resources.
@VoiceofSharif