AI models often identify civilians as military targets in tests
'simulated'
SUMMARY

— Advanced AI models repeatedly identified civilians and other protected people and sites as military targets in simulated scenarios, an independent research report found.

Advertisement

The mistaken targets included ambulances, surrendering soldiers and a school in Iran.

Anthropic PBC’s models performed worst. OpenAI’s models were among the best, but still made errors.

What do you think

Should AI be allowed to help select military targets?

One tap. No account.