arXiv:2603.11331v4 Announce Type: replace-cross Abstract: Adversarial attacks can reliably steer safety-aligned large language models toward unsafe behavior.
← Back to all developments · Signals are a development reported by more than one source. See the methodology.