Researchers have persuaded two popular Chinese artificial intelligence models to tell them how to construct biological weapons and carry out assassinations.
Mindgard, a company which tests the security of AI systems, found the Moonshot tools Kimi K2.6 and K3 Swarm could get round guardrails imposed by developers.
The discovery was made during a test known as ‘jailbreaking’, where researchers input detailed instructions to find out whether AI models ignore safety limits.
The systems gave advice on how to create sarin gas, generate malware software, take down planes and even plan a terrorist attack on the London Underground.
Once the model was jailbroken, the user
To provide well-rounded coverage and a breadth of insight across various events, we rely on contributions from several staff writers, each bringing their own area of expertise to our publication.