Radar Telegram alerts Open Radar

In testing, artificial intelligence models by Anthropic and others regularly mistook civilians for military targets”: AI mistook a school in Iran for a military target. "Advanced AI models routinely identified civilians, ambulances, surrendering soldiers and a school in Iran as military targets in simulated scenarios. The worst results were shown by models of Claude's chatbot developer, Anthropic. OpenAI's models were among the best, but still wrong. Multiple artificial intelligence models routinely identified the Minba school, hit by a deadly attack that killed more than 100 children on the first day of the major US battle earlier this year, as a military facility based on satellite images taken in 2025. One of Anthropic's models, Claude Opus 5.5, has more often than not chosen this school as a target.

תרגם לעברית

Пул N3 · · Source

🇮🇷 Iran · Map

In testing, artificial intelligence models by Anthropic and others regularly mistook civilians for military targets”: AI mistook a school in Iran for a military target. "Advanced AI models routinely identified civilians, ambulances, surrendering soldiers and a school in Iran as military targets in simulated scenarios. The worst results were shown by models of Claude's chatbot developer, Anthropic. OpenAI's models were among the best, but still wrong. Multiple artificial intelligence models routinely identified the Minba school, hit by a deadly attack that killed more than 100 children on the first day of the major US battle earlier this year, as a military facility based on satellite images taken in 2025. One of Anthropic's models, Claude Opus 5.5, has more often than not chosen this school as a target.

In testing, artificial intelligence models by Anthropic and others regularly mistook civilians for military targets”: AI mistook a school in Iran for a military target. "Advanced AI models routinely identified civilians, ambulances, surrendering soldiers and a school in Iran as military targets in simulated scenarios. The worst results were shown by models of Claude's chatbot developer, Anthropic. OpenAI's models were among the best, but still wrong. Multiple artificial intelligence models routinely identified the Minba school, hit by a deadly attack that killed more than 100 children on the first day of the major US battle earlier this year, as a military facility based on satellite images taken in 2025. One of Anthropic's models, Claude Opus 5.5, has more often than not chosen this school as a target.

More on Radar

Want to see every report in real time?

Hundreds of sources worldwide, translated into your language, with alerts for what matters to you.

Sign up free
Share on WhatsApp Get alerts on Telegram