Cointime reports: A new independent study released Nov. 17 found that advanced AI models frequently misidentified civilians, ambulances, surrendering soldiers and even a school in Iran as military targets in simulated scenarios. Anthropic’s model performed worst among those tested, while OpenAI’s performed relatively better but still made errors. The study, conducted by San Francisco-based AI evaluation firm Vals, tested models from eight companies and concluded that safeguards against misuse in autonomous weapons are scarce.
Study: Advanced AI Models Including Anthropic’s Frequently Misidentify Civilians as Military Targets
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.
Related News
58m ago
Microsoft launches six AI models to undercut OpenAI and Anthropic on cost
1h ago
OpenAI Launches GPT-6.2 Astra AI Model
4h ago
Anthropic IPO Draws Multiple $10B-Plus Subscriptions
5h ago
China’s Daily Token Usage Tops 500 Trillion as Compute Demand Surges
5h ago
Anthropic Said to Guide IPO Pricing at Top of Range or Above
5h ago
Anthropic IPO Oversubscribed 10-20x, Altimeter CEO Says