Control #

B

3

.

6

Evaluate AI against jailbreaks

Evaluate whether users can elicit restricted or harmful outputs via jailbreak strategies, including roleplay, obfuscation, indirect prompting, and chained inputs.

Evidence

We'll list specific evidence that demonstrates compliance with this control. Typically, this is screenshots, proof of a legal or operational policy, or product demonstrations.

Recommended actions

We'll recommend specific practices and actions for complying with this control.

Provide feedback on this control