AI Chatbot Allegedly Helped Couple Plan Triple Murder in Bengaluru

Police in the Indian city of Bengaluru say a couple accused of murdering three family members spent nearly six months using an AI-based chatbot to plan the crime step by step. Investigators found that the tool was allegedly asked how to dispose of the bodies, remove bloodstains, and whether a home oven could be used to destroy evidence.

Six Months of Chatbot Consultations

According to Indian police, Kenneth, an engineer and Shwetha's partner, repeatedly returned to conversations with an AI chatbot over six months, seeking answers related to the planned crime. The outlet Odishabytes reports the chatbot in question was Google Gemini, though some media accounts simply describe it as a popular AI chatbot without confirming the brand.

Investigators say Kenneth did not ask his questions directly. Rather than asking how to kill a specific person, he framed his queries hypothetically and indirectly, avoiding wording that might trigger the chatbot's safeguards.

If something like this were to happen, what should I do? - the question Kenneth allegedly put to the AI chatbot, according to Bengaluru police

What the Queries Looked Like

Police say the couple asked the chatbot, among other things, how to remove bloodstains from fabric and floors, how to incapacitate a victim before stabbing them, and what type of weapon would work best. Investigators also cited a query framed as a hypothetical question about defending against burglars, phrased in a manner similar to asking what weapon could be used to attack an intruder in the home.

Seeking answers to how to kill, and then how to dispose of the bodies, the couple used chatbots at every stage of the conspiracy - according to findings from the Bengaluru police investigation

An Oven Meant for the Kitchen

One of the most unsettling threads in the investigation involves an industrial oven that Kenneth had earlier built with the intention of launching his own cloud kitchen, a food business operating on delivery orders only. Police say the idea of using that same oven to burn the victims' bodies came up in conversation with the chatbot, which allegedly outlined both the advantages and drawbacks of the method before the bodies were hidden.

In the end, the bodies were not cremated in the oven, but the fact that a device built for another purpose was even considered as a tool for destroying evidence shows how far the planning went. Police stress that the couple consulted the chatbot at every successive stage of preparation, not just in a single moment of crisis.

Investigation and Demand for Chat Logs

After both suspects were taken into custody, police asked the company behind the chatbot to hand over the full history of conversations Kenneth conducted over those six months. This is meant to help reconstruct the timeline of the planning and determine to what extent the AI's responses actually shaped the course of the crime, versus the suspects merely seeking confirmation of intentions they had already formed.

Both Shwetha and Kenneth have been placed in remand custody and remain there. Investigators link the motive for the crime to strained family relations and a financial dispute over the loan Shwetha took from her mother to fund her partner's business.

Another High-Profile Case at the Intersection of AI and Crime

The Bengaluru case joins a growing list of incidents in which investigators around the world have pointed to AI chatbots playing a role in planning violence, from advising on suicide to offering guidance on harming others. Unlike earlier high-profile cases involving individual acts of suicide, the Indian case involves a premeditated triple murder planned over half a year.

The makers of popular chatbots, including Google, say they have built in safeguards that block responses to questions about harming others. This case, however, shows that questions phrased indirectly or hypothetically can slip past such filters, especially when a conversation stretches across many months and many separate threads. For readers in Poland, the case carries a regulatory signal: the European Union is currently rolling out further provisions of the AI Act covering the obligations of providers of the most powerful models, and cases like Bengaluru's are likely to be cited in debates over how effective current safeguards really are.

The investigation in Bengaluru is ongoing, and police have not yet disclosed whether the chatbot conversation logs will be used as evidence in the formal charges. The case could become a precedent if an Indian court decides to treat a history of AI conversations as part of the evidentiary chain establishing premeditation.