DiffusionGemma matches Gemma 4 in variable transparency and monitorability after applying an interpretable token bottleneck, despite higher naive serial depth, and shows novel phenomena such as non-chronological reasoning.
arXiv preprint arXiv:2501.17315 , year=
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
years
2026 6verdicts
UNVERDICTED 6roles
background 1polarities
background 1representative citing papers
Lie detectors effective on prompted deception in LLMs fail on trained model organisms with verified opposite beliefs, except chain-of-thought judges which retain 0.82 balanced accuracy partly due to verification artifacts.
Strategic attack selection via start and stop policies reduces empirical safety by 20-28pp in BashArena and LinuxArena agentic control evaluations without changing attack capability.
Diverse ensembles of prompted and fine-tuned GPT-4.1-Mini monitors achieve 2.4x better detection of flawed code solutions than homogeneous ensembles on adversarial inputs.
Behavioral assurance is structurally unable to verify the latent safety properties demanded by AI governance frameworks enacted 2019-2026.
ATLAS shows constitutions induce recoverable latent geometry in LLMs that redistributes but remains detectable across models and neural perturbation data via source-defined families and AUC separations.
citing papers explorer
-
How Transparent is DiffusionGemma?
DiffusionGemma matches Gemma 4 in variable transparency and monitorability after applying an interpretable token bottleneck, despite higher naive serial depth, and shows novel phenomena such as non-chronological reasoning.
-
"Did you lie?" Evaluating Lie Detectors across Model Scale and Belief-Verified Model Organisms
Lie detectors effective on prompted deception in LLMs fail on trained model organisms with verified opposite beliefs, except chain-of-thought judges which retain 0.82 balanced accuracy partly due to verification artifacts.
-
Attack Selection in Agentic AI Control Evaluations Meaningfully Decreases Safety
Strategic attack selection via start and stop policies reduces empirical safety by 20-28pp in BashArena and LinuxArena agentic control evaluations without changing attack capability.
-
Ensemble Monitoring for AI Control: Diverse Signals Outweigh More Compute
Diverse ensembles of prompted and fine-tuned GPT-4.1-Mini monitors achieve 2.4x better detection of flawed code solutions than homogeneous ensembles on adversarial inputs.
-
Position: Behavioural Assurance Cannot Verify the Safety Claims Governance Now Demands
Behavioral assurance is structurally unable to verify the latent safety properties demanded by AI governance frameworks enacted 2019-2026.
-
ATLAS: Constitution-Conditioned Latent Geometry and Redistribution Across Language Models and Neural Perturbation Data
ATLAS shows constitutions induce recoverable latent geometry in LLMs that redistributes but remains detectable across models and neural perturbation data via source-defined families and AUC separations.