A critical security flaw in AI coding tools allows attackers to bypass sandbox environments by misleading human reviewers.