Weaknesses Found in AI Safety Testing: New Insights

Experts have discovered flaws in AI safety tests that could compromise the validity of their results.