Study Reveals Safety Gaps in AI-Controlled Robots
A study by the Robocurve foundation found that manipulating images fed to AI models controlling robots pushed them to attempt dangerous tasks in up to 97% of tests.
Safety
Everything tagged Safety.
Everything tagged Safety.
A study by the Robocurve foundation found that manipulating images fed to AI models controlling robots pushed them to attempt dangerous tasks in up to 97% of tests.