Poisoning cases have their own evidentiary, mens rea, and causation issues to be established with reasonable certainty. These difficulties are more evident in cases concerning new psychoactive substances (NPS) and herbal poisonings such as Abrus precatorius and Nerium oleander, where detection is difficult, untimely, and often inconclusive. Standard toxicological methods cannot detect new chemicals or properly quantify herbal poisonings to prove their use with the intent to kill. This study will examine the potential of Artificial Intelligence (AI) in toxicological screenings in poisoning cases. AI models can detect and identify unknown substances and infer patterns. This study is crucial for examining whether such innovations could help fill the evidentiary gap in mens rea or raise any legal concerns regarding admissibility and reliability. In light of the merging nature between forensic science and criminal law, the paper suggests that there is much room for AI-assisted toxicology to strengthen the prosecution’s case, yet whether this approach can legally be applied at the moment remains questionable. Moreover, if regulations remain unrigorous and judicial practices are lenient, it could prove less compelling than other cases in arguing for its viability. Ultimately, the author believes that an appropriate balance needs to be struck whereby the technological advancements align with the core tenets of criminal law. The study further undertakes a comparative analysis of regulatory approaches adopted in the United States, the European Union, the United Kingdom, and Australia to identify best practices governing AI-assisted forensic evidence. Based on these findings, it proposes legislative and procedural reforms for India, including clear admissibility standards, validation protocols, transparency requirements, and specialised regulatory oversight to ensure that AI-assisted forensic toxicology strengthens criminal investigations while safeguarding the rights of the accused.