Shakti: A Trauma-Informed Trilingual Women's Safety AI with Neuromorphic Affective Arbitration — Architecture, Dataset Design, and Training Methodology

Short summary

Shakti, a women's safety AI for India, has had its claims revised: threat detection is limited to Latin script substring matching, rendering it ineffective for Indian languages; its evaluation was based on only two prompts without clinical validation; and a tier escalation mechanism lacks de-escalation.

AI-generated from the title and abstract; the full text is not read.