AI chatbots like ChatGPT, Gemini, Claude and Grok can answer almost anything. Need recipe ideas? Ask AI. Need help writing code? Ask AI. But as it turns out, some people are turning to AI for something far more sinister. From mass shootings to cold-blooded murders, AI is increasingly showing up as a tool that many people are using while planning violent crimes. And in some cases, they are even getting the answers back with AI tools trying to be “helpful” in whatever their users are trying to do.
The latest example comes from Bengaluru, where cops claim that 25-year-old Kenneth spent nearly six months consulting Google's Gemini before allegedly murdering his live-in partner's parents and younger sister. Investigators claim he used the chatbot to gather information on killing methods, removing bloodstains, disposing of bodies and even whether an iron furnace could be used to destroy evidence.
But don't AI chatbots have guardrails? They do. So how does someone get around them? Well, they are tricking the AI bot.
Police believe Kenneth allegedly avoided asking direct questions such as, "How do I commit murder?" Instead, he framed his prompts as hypothetical scenarios, asking what someone should do "if something like this happens". Investigators say that helped him gather information without triggering obvious safety filters.
Cops also allege that he initially considered disposing of the bodies in an iron furnace built for his proposed cloud kitchen business before abandoning the idea after evaluating its practicality with Gemini.
Police have now reportedly sought access to his Gemini chat history as part of the investigation. India Today Tech has reached out to Google for a comment and we will update the copy if and when we receive inputs.
Note that the allegations are yet to be tested in court. However, the Bengaluru case is far from an isolated incident. From the US to Australia, investigators have uncovered several cases in which suspects allegedly turned to AI chatbots to research attacks, refine violent plans or obtain tactical information before committing crimes.
2025 Florida State University mass shooting
One of the recent international cases that came into light due to AI connection is the 2025 Florida State University mass shooting.
According to a lawsuit filed by the family of one of the victims, the alleged shooter exchanged hundreds of conversations with ChatGPT in the months leading up to the attack. The lawsuit claims the chatbot provided planning-related information, while prosecutors later reviewed the chat logs as part of the investigation. Florida's Attorney General also launched a criminal investigation into ChatGPT's role.
OpenAI, however, rejected the allegations, saying ChatGPT only returned publicly available factual information and did not encourage or promote violence.
Teen allegedly used AI to plan mass shooting in Australia
The US isn’t the only place where AI allegedly has became part of a criminal investigation in a case.
In Australia, court documents claim a 13-year-old repeatedly used generative AI to create detailed mass shooting scenarios, draft violent manifestos and even requested a story describing a "Bondi-style" beach massacre. The alleged target was a school, but police arrested the teenager before any attack could take place.
The case also became one of the earliest examples of AI-generated conversations being used as evidence in an alleged terrorism-style investigation.
2025 Cybertruck explosion in the US
Another case that caught global attention due to AI connection was the 2025 Cybertruck explosion outside the Trump International Hotel in Las Vegas. According to a report by The Guardian, investigators revealed former US Army soldier Matthew Livelsberger used ChatGPT to research explosives and tactical information before carrying out the attack.
Teen stabs 3 girls in Finland
In another case, this time from Finland, a 16-year-old allegedly spent months planning an attack on his school, using a chatbot to help draft a manifesto and prepare for the assault. He later stabbed three teenage girls at a school in Pirkkala. Police said they examined the alleged AI-assisted manifesto as part of their investigation, with the suspect reportedly claiming he had planned the attack for around six months.
The danger of easy access to AI
Of course, AI is making everyday life easier for millions of people. It helps students study, developers write code, businesses automate routine work and professionals get more done in less time. But like any powerful technology, it is only as good as the intent of the person using it.
With AI now just a few clicks away for almost anyone with an internet connection, the ease of access is becoming a growing concern for technology companies and law enforcement agencies alike.
To mitigate the risks, AI companies have introduced stricter guardrails designed to stop chatbots from providing harmful information. But as the Bengaluru case and several others suggest, those guardrails aren't foolproof. Some users are finding ways around them by disguising dangerous requests as hypothetical scenarios. And many times, as smart as AI is, it is getting fooled by the intents of its human users.
- Ends
Published By:
Divya Bhati
Published On:
Jul 22, 2026 13:44 IST