A Chinese AI tool was able to tell researchers how to carry out assassinations and make biological weapons. Developer Moonshot is conducting an internal review after researchers were able to persuade two of its popular Kimi models to evade its safety limits. This took place during a process called "jailbreaking", where researchers use a series of complex instructions to see if AI tools ignore guardrails. Tap the link in @BBCNews’s bio to find out how the company responded. #Kimi #AI #BBCNews