Researchers at cybersecurity firm Tracebit have developed context bombing, a new AI defence that uses prompt injections to disrupt AI-powered hacking agents by triggering their own safety guardrails.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results