F-Secure maps 26 ways AI can harm consumers

Los Angeles, 10 November 2026 — F-Secure has released an AI Harms Taxonomy developed by Dr. Megan Squire, Principal Threat Intelligence Researcher at F-Secure. The framework identifies 26 distinct harms faced by consumers as AI becomes more embedded in search, communication, decision-making and daily digital life. It traces how harm travels through AI systems and across the services they rely on, pairing each potential impact with where protections can intervene and who is best placed to act. The taxonomy emphasises that risk now often involves AI as an intermediary in user interactions, rather than direct attacks on individuals.

The research maps 26 harms across five relationship types and 19 intermediary layers, with 24 defenses designed to address risks at different points. Concrete examples include AI-generated voice and video impersonation, agents making harmful financial decisions, AI exposing personal information or acting on outdated context, and scenarios where malicious instructions are hidden in content processed by an AI.

The study also stresses that responsibility for preventing AI harm is distributed among several parties: some protections can sit on a consumer’s device, but others must be implemented further upstream by the companies building AI products and the systems they connect to. The work forms part of F‑Secure’s Trust initiative, which includes TrustGuard and TrustPath, and is intended to underpin practical safeguards as AI agents increasingly act on behalf of users.