Two Israelis Indicted for Selling AI-Faked Secrets to Iran

Two Israeli brothers have been indicted on serious security charges after allegedly maintaining contact with an Iranian agent and providing him with fabricated intelligence generated using artificial intelligence tools, authorities said Tuesday.

The indictment, filed by the State Attorney’s Office following a weeks-long press embargo, accuses the brothers, residents of Beitar Illit and Beit Shemesh, of contact with a foreign agent and passing information to the enemy.

According to prosecutors, the pair received more than NIS 100,000 from the Iranian handler in exchange for information, much of which was entirely fabricated.

The main defendant is accused of initiating and maintaining the relationship after being contacted via Telegram in August by an individual offering payment for information. Suspecting the contact was an Iranian operative, he allegedly chose to engage while presenting himself under a false identity.

Prosecutors say the defendant created an elaborate deception, claiming to be a computer science student with connections to the IDF’s Unit 8200. He allegedly fabricated a fictional soldier, using real personal details found online, and staged conversations and documents to convince the handler of his credibility.

The indictment details how the defendant used artificial intelligence platforms, including ChatGPT, Grok, and Gemini, to generate false intelligence reports. These included claims about a future Israeli-American strike on Iran, as well as lists of strategic targets compiled using publicly available tools such as Google Maps.

In one instance, he allegedly created a forged document accusing an Iranian civilian of collaborating with Israel. The individual was later arrested by Iranian authorities, interrogated, and ultimately cleared of suspicion.

The defendant is also accused of fabricating responses to questions about Unit 8200 operations, generating documents and images, including a falsified photograph of the supposed soldier holding identification, using AI tools.

According to the indictment, some information passed to the Iranian agent was based on real conversations the defendant overheard or interpreted, including speculation about potential Israeli military activity. Prosecutors allege he relayed these claims as confirmed intelligence.

The case raises concerns about the use of widely available AI tools to produce seemingly credible but false intelligence, as well as the risks posed by direct contact with hostile foreign agents, regardless of intent.

The brothers’ defense attorney rejected the charges, arguing that his clients acted out of patriotism and sought to deceive the Iranian agent rather than assist him.

“These are two patriotic individuals who intended to mislead the Iranian enemy,” the lawyer said. “Instead of prosecuting them, their actions should be recognized for contributing to national security.”

Authorities have not accepted that argument, and the case is proceeding under charges typically associated with espionage-related offenses.

No trial date has been announced.