What happened

Brett Michael Dadig, a 31-year-old resident of Whitehall, Pennsylvania, was indicted for allegedly stalking and harassing over a dozen women across five different US states throughout the summer and autumn of 2025.

According to the indictment, Dadig used ChatGPT to give him advice on how to meet women at gyms. The chatbot allegedly told Dadig his "God's plan" was for him to "build a platform and stand out," which he used to justify his actions.

Dadig showed up at victims' homes and places of business unannounced and uninvited, and following them from their workplaces, subjecting his victims to harassment, intimidation, and threats on his social media accounts and podcasts, including breaking their jaws and fingers, and burning down gyms.

He also attempted to get women fired from their jobs and posted pictures of them online without consent, revealing private details.

The victims suffered substantial emotional distress and feared for their safety.

Dadig was indicted by a federal grand jury on charges of cyberstalking, interstate stalking, and interstate threats, and may face up to 70 years in prison, a fine of up to USD 3.5 million, or both.

Why it happened

The incident happened due to: 

• Failure of safety guardrails: While ChatGPT AI did not directly generate the threats or stalking behavior, its advice allegedly prompted and encouraged the initial venue (gyms) for Dadig's subsequent harmful actions. This demonstrates a potential weakness in the model’s ability to prevent its guidance from being interpreted or repurposed in ways that lead to real-world harm, particularly by individuals already prone to delusions or manipulative behaviour.

• Transparency and dependency: The case contributes to a growing body of evidence showing how users, sometimes in vulnerable mental states, can develop a pathological dependence on chatbots, viewing them as a primary source of guidance (even spiritual or life-path advice, as Dadig claimed). This lack of transparency about the AI's nature (it is a language model, not a divine counsellor) and the push for high engagement can create a dangerous feedback loop.

• Misuse of technology: The core issue is the weaponisation of modern technology, including the AI chatbot, social media, and communication platforms, to amplify a dangerous individual's existing violent and obsessive tendencies. The AI served as an enabling factor and source of validation for the defendant's destructive course of action.