AINewsnow

AI Models Regularly Identify Civilians as Military Targets in Simulated Warfare Tests

An independent evaluation by Vals found AI models from Anthropic, Google, xAI, DeepSeek, and OpenAI misidentified civilians, ambulances, surrendering soldiers, and a school in Iran as military targets in simulated scenarios, and could be manipulated by framing the task as a game.

Read the full story at Bloomberg AI ↗

Timeline · 5 reports

  1. 2026-11-17 17:20 · TechCrunch AI
    AI models can be tricked into identifying civilians as military targets by telling them it's a game
  2. 2026-11-17 17:10 · Techmeme
    Vals: in simulated scenarios, AI models from Anthropic, Google, xAI, DeepSeek, and OpenAI identified civilians, ambulances, and a school as military targets
  3. 2026-11-17 16:10 · The Next Web
    AI models flagged civilians as military targets in tests, Bloomberg reports
  4. 2026-11-17 15:00 · Bloomberg AI
    Top AI Models Regularly Identify Civilians as Military Targets in Tests
  5. 2026-11-17 15:00 · Bloomberg AI
    Anthropic, Other AI Models Regularly Identify Civilians as Military Targets in Tests

More stories

  1. Do AI chat engines block scrapers, and how do providers get around it? — DEV Community — AI
  2. Turn Free Web AIs (DeepSeek, Gemini, ChatGPT, Claude) into a Local MCP Agent Harness — DEV Community — AI
  3. Coding agent ecosystem matures as developers run Claude Code, Codex and Gemini CLI side by side — DEV Community — AI
  4. Claude, ChatGPT, Gemini, and Grok produced viable terror plans in test, per GAO — FedScoop
  5. DeepSeek vs Claude vs Kimi vs Codex vs Gemini: A 70-Call AI Ethics Test — DEV Community — AI
  6. How to prepare for an OpenAI interview in 2026: comparing the main practice options — DEV Community — Machine Learning
  7. ChatGPT, Claude, Gemini and Grok helped plan terror attacks in US government test — The Independent Tech
  8. Perplexity Launches Orchestrator API With Same-Model Failover — Unite.AI

Read the Daily Brief: the day's most important AI stories →