F-Secure Maps 26 Ways AI Can Harm Consumers and Where Defences Can Step In
Security firm F-Secure has published a new AI Harms Taxonomy that identifies 26 ways artificial intelligence can harm consumers. The framework follows how harm moves through AI systems, covering a wide range of problems from misleading chatbot answers through to compromised autonomous agents.
A key part of the work is identifying the points where defences can intervene. Rather than treating AI risk as one large, vague problem, the taxonomy breaks it into stages, so that organisations can see where a safeguard might stop harm before it reaches a person.
For Australian small businesses, the report is a reminder that AI risk is not limited to one type of tool. Customer-facing chatbots that give poor answers and AI agents that act on a user's behalf can both cause harm, and the second case becomes more serious if the agent is compromised. The source summary does not go into further detail, so businesses should consult the full report from F-Secure for the specific harms and recommended controls.