Threat · curated 13 Jul 2026
EchoGram and guardrail bypass: are AI defenses keeping up?
First reported · updated · 2 reports hiddenlayer.com
Coverage timeline
Why it matters
EchoGram shows that the guardrail models many teams trust to protect LLMs and AI agents can be silently bypassed with adversarial token suffixes, undermining a core AI safety control.
HiddenLayer research dubbed EchoGram demonstrates that carefully chosen token sequences can flip verdicts in LLM guardrail models, causing harmful prompts to be marked safe or benign prompts to trigger false alarms. The NHIMG editorial summarizes the finding and its implications for organizations relying on probabilistic AI safety layers to protect deployed LLMs and agents.