5 min read
Research Check Point Research found that when teammates reversed on a rule, a frontier-model agent opened a forbidden folder in up to 98% of runs, up from 44%. The agent almost never said why.
Check Point Research found that when teammates reversed on a rule, a frontier-model agent opened a forbidden folder in up to 98% of runs, up from 44%. The agent almost never said why.
RAG pipelines introduce document poisoning, indirect prompt injection via retrieved content, and semantic access control gaps that most security teams have not assessed.