Digital pathways to self-harm: Modafinil-induced mania and AI-assisted operationalization of suicidal intent in a law student: a case report

Abstract

Background: Modafinil is a wakefulness-promoting (eugeroic) agent increasingly used off-label for cognitive enhancement. Although generally regarded as having a favorable psychiatric safety profile, it can precipitate manic or hypomanic episodes in vulnerable individuals, and the boundary between a self-limited substance-induced episode and the unmasking of an underlying mood disorder is not always clear at presentation.

Case Presentation: We report the case of a 20-year-old female law student admitted to the emergency unit after a suicide attempt by polypharmacy overdose during escalating off-label modafinil use. A notable feature was her use of an artificial intelligence (AI) conversational platform to inform about the preparation of the attempt: the AI did not generate her suicidal motivation but contributed to operationalizing an already-formed intent. The acute presentation was characterized by logorhoea, affective expansiveness, and behavioral disinhibition, initially consistent with a substance-induced manic episode. Over the three-month follow-up, however, subsyndromal hypomanic features persisted after discontinuation of modafinil and after withdrawal of a serotonergic antidepressant, while the patient was maintained on a mood stabilizer — raising the possibility that modafinil acted as a precipitant unmasking an underlying bipolar diathesis.

Conclusions: This case illustrates modafinil’s potential to precipitate affective destabilization in a high-achieving young person under sustained academic pressure, and the difficulty of distinguishing a transient substance-induced episode from an emerging bipolar-spectrum disorder without longitudinal follow-up. It also highlights an emerging risk: the role of large language models (LLMs) in operationalizing — rather than originating — suicidal intent. The case underscores the need for careful psychiatric screening before prescribing wakefulness-promoting agents, for continued longitudinal monitoring, and for attention to patients’ interactions with AI platforms during crisis assessment.