SpecGuard monitors multi-turn LLM jailbreaks with temporal logic
A training-free monitor compiles Linear Temporal Logic rules into automata and emits replayable violation witnesses.
Stories per day · 30 days
3 RSS
Relatedlanguage-models1llm-safety1pretraining-seeds1reproducibility1runtime-monitoring1specguard1temporal-logic1
5 stories, newest first · one card per story
Stories tagged #jailbreaks
Pretraining seeds affect jailbreak success more than data recipes in language models
Identical pretraining runs with different seeds produce models with varying susceptibility to 91 jailbreak templates.
Vani contrasts derisked, uncensored and jailbroken models
The post says derisked models drop the consumer refusal layer while keeping base capability and terms.
1
We’re putting too much faith in AI’s ability to say no
MIT Technology Review argues AI safety leans too heavily on probabilistic, company-defined refusals that can fail or censor.
TIER: Threat Implicitness Benchmark for Evaluating LLM Safety Behaviors
TIER benchmark shows LLM safety behaviors shift gradually across threat implicitness levels, with jailbreaks exposing the largest robustness gaps.