AI models from eight developers marked civilians, ambulances, surrendering soldiers and a school as military targets during tests by Vals, a San Francisco-based AI evaluator. Researchers Geby Jaff and Anthony Ozerov examined how readily the models would provide information that could be used by autonomous weapons systems. They did not ask the models to authorize […]
The post Anthropic, other AI models flag civilians as military targets in tests appeared first on Insider Paper.