Threat Labs

Tenet Threat Labs has demonstrated a new class of attack "Agentjacking" that hijacks AI coding agents into running attacker-controlled code on a developer's machine, triggered by a single fake error report and invisible to every security control.

AI agents are being deployed faster than the security controls to manage them. Three questions every security leader should be able to answer before any agent ships: what it can do, what controls exist if it misbehaves, and how you'll know when something goes wrong.

Tenet Threat Labs recently captured three live attacks targeting enterprise AI agents — prompt injection, CoT goal manipulation, and MCP-layer exploitation. None were flagged by conventional tools. Active exploitation isn't coming - it's already here.

Tenet Research declassifies a critical S3/CDN flaw in Hugging Face that exposed private LLM weights via a "header-override" exploit. This case proves that static configs aren't enough. Runtime defense is the only way to secure the agentic baseline. Full teardown here:

Enterprise AI has evolved. We have moved past simple chat interfaces to autonomous agents with the power to query databases, call APIs, and execute code. But this autonomy comes with a trade-off: a new breed of sophisticated, multi-stage attacks that turn an agent's reasoning against itself.