Saleh’s forces alleged that an AI-generated audio resembling their commander’s voice was circulated during the battle for the strategic port city of Mocha.
The Iran-backed Houthi forces have been accused of using artificial intelligence to create an audio recording resembling the voice of Yemeni military commander Lieutenant General Tareq Saleh. The alleged AI-generated recording reportedly circulated during the Houthis’ offensive to capture the strategic port city of Mocha.
Saleh’s National Resistance Forces alleged that the Houthis used artificial intelligence to clone the commander’s voice and circulate a message calling on troops to withdraw. The alleged audio was described by Saleh’s forces as an attempt to create confusion and weaken the morale of soldiers during intense fighting.
The Houthis have seized the strategic Red Sea city of Mocha, expanding their presence near the Bab al-Mandeb Strait, one of the world’s most important shipping routes. The Iran-backed group captured the city on Thursday after forces of the internationally recognised government withdrew, according to sources cited by Al Jazeera.
AI Audio Allegedly Told Troops To Withdraw
The disputed audio recording reportedly featured a voice resembling Saleh and included a message to withdraw from Mocha, according to the Basha Report, a risk advisory firm.
The audio, circulated in Saleh’s name in Arabic, said, “Heroes, these young men have learnt that the tide can turn in battle. War demands both advance and retreat. Some days favour you, and others do not, and there is no shame in that. We have reserves, additional strength and the men to continue fighting.”
According to Saleh’s National Resistance Forces, the recording was fabricated using artificial intelligence and falsely attributed to the commander. Mocha-based satellite television channel Aljoumhouriya TV also reported on the alleged AI audio and described it as an attempt to mislead public opinion and undermine troop morale.
“Military Media: The Houthis fabricated an audio recording using artificial intelligence and attributed it to Major General Staff Tareq Saleh in a failed attempt to mislead public opinion and affect morale,” it posted on X.
Saleh’s media arm later said that his official statements would only be released through authorised National Resistance military channels and warned people against trusting unauthenticated messages. The reported audio circulated as Houthi fighters advanced through positions held by Saleh’s forces along Yemen’s western coast.
Messages And Audio Allegedly Used To Create Confusion
The alleged AI-generated recording was reportedly part of a broader attempt to confuse government-aligned troops during the fighting. According to reports, Houthi forces also sent messages offering amnesty to soldiers who surrendered their weapons and returned home, according to the Basha Report.
The alleged audio recording and messages appeared aimed at convincing troops that resistance was weakening and that withdrawal or surrender was their safest option. However, the reported messages and their impact on soldiers have not been independently verified.
AI Allegation Emerges Alongside Claude Report
The allegation of AI-generated audio emerged alongside separate reports of possible AI use by actors operating in Houthi-controlled areas. Anthropic recently released a threat intelligence report detailing how governments and non-government groups had attempted to use AI models for activities linked to warfare.
The report described efforts by a weapons-development group in northern Yemen to use Claude Code for work related to guidance, navigation and control software for missile and rocket programmes. It banned the accounts involved and “shared threat information with public- and private-sector partners to mitigate risks posed by the actors.”
