Google analytics are off.

Promptective journal / Topic index

AI safety

1 journal entry tagged AI safety.

Cream and dark instruction strips take different routes through a slatted model enclosure, with a cobalt clamp closing the side gap

Guide9 February 20263 min read

How LLM jailbreaks bend the rules

Techniques, risks and practical defences.

How attackers bypass LLM safety guardrails, the common techniques they use, and practical ways to defend AI applications.

Read guide

All Articles, Guides and News