Amazon's recommendation algorithm was discovered to be guiding users to purchase chemical combinations for producing explosives and incendiary devices by suggesting explosive ingredients together in 'Frequently bought together' sections.
Channel 4 News discovered that Amazon's product recommendation algorithm was suggesting chemical combinations that could be used to produce explosives and incendiary devices. The algorithm grouped ingredients for black powder and thermite together under 'Frequently bought together' sections on chemical product listings. Users searching for common chemical compounds used in food production were offered ingredients to produce explosive black powder, with Channel 4 News able to create a shopping basket containing up to 45kg of black powder ingredients. Similarly, users searching for another widely available chemical were offered other thermite ingredients in recommendation sections. The algorithm also suggested steel ball bearings, push button switches, battery connectors, cables, igniter cord, and electronic ignition systems alongside these chemicals. One chemical being sold was listed as a 'Regulated substance' by the Home Office, making it illegal to possess without a licence. While many individual ingredients are not illegal to buy or sell in the UK, there have been successful prosecutions where people bought multiple chemicals and electronic components necessary to produce explosives. Amazon stated that all products must adhere to their selling guidelines and UK laws, and that they work closely with police and law enforcement agencies when needed.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to develop cyber weapons (e.g., by coding cheaper, more effective malware), develop new or enhance existing weapons (e.g., Lethal Autonomous Weapons or chemical, biological, radiological, nuclear, and high-yield explosives), or use weapons to cause mass harm.
AI system
Due to a decision or action made by an AI system
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
No population impact data reported.