I Broke Claude Opus 5 With Three Words. Here’s What That Means.
A three-word prompt broke Claude Opus 5, the most advanced AI model. This isn’t just a bug—it exposes a fundamental flaw: safety filters are built on surface-level patterns, not deep understanding. If a trivial phrase can bypass billions in safety research, then AI alignment is a mirage, and every trust placed in these systems is fragile.