23965 shaares
AI agents are indifferent to boundary trampling, and cannot observe ethics. They pretty routinely break out of prompt guardrails; this isn't even a new thing.
Interesting contextual integrity violation.