Researchers Bypass Safeguards on Moonshot’s Kimi AI to Generate Harmful Content
This event underscores AI safety vulnerabilities that could be exploited to disseminate dangerous information on biological weapons and assassination methods.
This event underscores AI safety vulnerabilities that could be exploited to disseminate dangerous information on biological weapons and assassination methods.
Auto-scored: Demonstrated breach of AI safeguards constituting real escalation through misuse.