Threat · curated 2 Oct 2026
Chinese AI tool told researchers how to make bioweapons
First reported bbc.com
Coverage timeline
Single-source incident — first reported, latest, and curated coincide.
Why it matters
Jailbreaks of widely deployed LLMs like Kimi can strip safety guardrails and turn a public model into a source of weapons guidance or a code-execution launchpad for attacks, showing provider safety limits remain bypassable.
Security firm Mindgard reported it jailbroke Moonshot's Kimi K2.6 and K3 Swarm models, bypassing safety guardrails to elicit instructions for making biological weapons and carrying out assassinations. Mindgard also said a jailbroken Kimi 2.6 could run code and connect to the internet, potentially serving as a launchpad for cyber-attacks. Moonshot said it is conducting an internal review and in discussion with Mindgard.