Connect OpenTG to ChatGPT or Claude in 60 seconds

Try it

A new independent study found that advanced AI models frequently misidentified civilians, ambulances

Clash Report

Clash Report

@ClashReport

Breaking news, reports, and opinions from ongoing clashes of the world. https://www.instagram.com/clashreportcr https://x.com/clashreport

110,149 subscribers
Open in Telegram
A new independent study found that advanced AI models frequently misidentified civilians, ambulances, surrendering soldiers and a school as military targets in simulated scenarios.

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

Source: Bloomberg
Open post in Telegram