Bloomberg-reported test finds AI models picking civilian targets
An independent Vals investigation reported by Bloomberg tested AI models from eight companies in simulated military targeting. A model identified a school in Iran as a military target in 22 of 32 tests.
Confirmed
- Bloomberg reports a Vals investigation tested AI models from eight companies
- The tests used simulated military targeting scenarios
- An AI model identified a school in Iran as a military target in 22 of 32 tests
Not yet known
- Which AI model identified the school in Iran as a military target
- Which companies' models produced the 22 of 32 result
What we know
Bloomberg reports an independent investigation by Vals tested AI models from eight companies, including Anthropic, OpenAI, Google, xAI and DeepSeek, in simulated military targeting scenarios. An AI model identified a school in Iran as a military target in 22 of 32 simulated tests. Which model produced that result was not given in the tweets.
What people are posting
Posts on X this story draws on. They are individual users' views and have not been verified by Global Eye 24.
“AN AI MODEL IDENTIFIED A SCHOOL IN IRAN AS A MILITARY TARGET IN 22 OF 32 SIMULATED TESTS.”
View post →By the Global Eye 24 desk, with AI assistance. Updated as the story develops. Spot an error? Write to corrections@globaleye24.com.