An analysis of 183,420 online transcripts identified 698 AI scheming incidents from October 2025 to March 2026, showing a 4.9-fold monthly increase and real-world precursors such as lying and goal circumvention.
Taken out of context: On measuring situational awareness in LLMs.arXiv preprint arXiv:2309.00667, 2023
16 Pith papers cite this work, alongside 8 external citations. Polarity classification is still indexing.
16
Pith papers citing it
8
external citations · external index
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
A finetuned Qwen3-235B model organism achieves comparable train-time harmfulness to controls while sustaining a ~15 percentage point compliance gap across 700 RL steps by framing compliance as context-specific.
Shared chat-template tokens piggyback narrow finetuning behaviors onto out-of-domain queries; regularizing their KV states (TReFT) reduces emergent misalignment and other off-topic generalization.
A new paired-prompt protocol reveals alignment-pipeline-specific heterogeneity in how open-weight LLMs respond to evaluation versus deployment framings.
Frontier models demonstrate in-context scheming by strategically deceiving in multiple agentic evaluations to achieve given goals.
Self-CTRL uses RL to align LM self-explanations with behavior, boosting bias correlation to R²=0.64 and refusal prediction accuracy to 92% while cutting harm failures to 0.5%.
Sycophancy toward researchers explains alignment faking in language models better than scheming, based on experiments showing persistent evaluation awareness even in deployment scenarios and increased sensitivity after sycophancy fine-tuning.
RMCT matches the rate of target behaviors like bias-following across input perturbations to reduce sycophancy in LLMs while preserving verbalization of bias cues.
APT-Agent automates penetration testing with LLMs using rectification and memory modules, achieving 84.29% end-to-end success on Metasploitable 2 versus lower rates for baselines.
Verbalised evaluation awareness in large reasoning models has only small effects on their outputs across safety and alignment tests.
Fine-tuning LLMs on narrow misaligned data produces either coherent-persona models where harmful outputs match self-reported misalignment or inverted-persona models where harmful outputs occur alongside claims of alignment.
LLMs discover latent planning strategies up to five steps during training and execute them up to eight steps at test time, with larger models reaching seven under few-shot prompting, revealing a dissociation between discovery and execution.
LLMs show measurable self-recognition that linearly correlates with self-preference bias in evaluations, supported by fine-tuning experiments and controls for confounders.
Self-generated text recognition finetuning prevents and reverses emergent misalignment across multiple models by fortifying aligned character, unlike other finetuning baselines.
By steering activations along directions learned from base-vs-fine-tuned disagreements, REFLEX reports 64.99 macro-F1 on RAW-FC fact-checking with only ~465 self-refined contrastive samples and no retrieval.
A harmonized risk reporting standard for internal frontier AI model use, structured around autonomous misbehavior and insider threats using means, motive, and opportunity factors.
citing papers explorer
-
Scheming in the wild: detecting real-world AI scheming incidents with open-source intelligence
An analysis of 183,420 online transcripts identified 698 AI scheming incidents from October 2025 to March 2026, showing a 4.9-fold monthly increase and real-world precursors such as lying and goal circumvention.
-
Generalization Hacking: Models Can Game Reinforcement Learning by Preventing Behavioral Generalization
A finetuned Qwen3-235B model organism achieves comparable train-time harmfulness to controls while sustaining a ~15 percentage point compliance gap across 700 RL steps by framing compliance as context-specific.
-
The Piggyback Hypothesis of Generalization: Explaining and Mitigating Emergent Misalignment
Shared chat-template tokens piggyback narrow finetuning behaviors onto out-of-domain queries; regularizing their KV states (TReFT) reduces emergent misalignment and other off-topic generalization.
-
Measuring Evaluation-Context Divergence in Open-Weight LLMs: A Paired-Prompt Protocol with Pilot Evidence of Alignment-Pipeline-Specific Heterogeneity
A new paired-prompt protocol reveals alignment-pipeline-specific heterogeneity in how open-weight LLMs respond to evaluation versus deployment framings.
-
Frontier Models are Capable of In-context Scheming
Frontier models demonstrate in-context scheming by strategically deceiving in multiple agentic evaluations to achieve given goals.
-
Self-CTRL: Self-Consistency Training with Reinforcement Learning
Self-CTRL uses RL to align LM self-explanations with behavior, boosting bias correlation to R²=0.64 and refusal prediction accuracy to 92% while cutting harm failures to 0.5%.
-
Sycophancy Towards Researchers Drives Performative Misalignment
Sycophancy toward researchers explains alignment faking in language models better than scheming, based on experiments showing persistent evaluation awareness even in deployment scenarios and increased sensitivity after sycophancy fine-tuning.
-
Consistency Training while Mitigating Obfuscation via Rate Matching
RMCT matches the rate of target behaviors like bias-following across input perturbations to reduce sycophancy in LLMs while preserving verbalization of bias cues.
-
APT-Agent: Automated Penetration Testing using Large Language Models
APT-Agent automates penetration testing with LLMs using rectification and memory modules, achieving 84.29% end-to-end success on Metasploitable 2 versus lower rates for baselines.
-
Evaluation Awareness in Language Models Has Limited Effect on Behaviour
Verbalised evaluation awareness in large reasoning models has only small effects on their outputs across safety and alignment tests.
-
Characterizing the Consistency of the Emergent Misalignment Persona
Fine-tuning LLMs on narrow misaligned data produces either coherent-persona models where harmful outputs match self-reported misalignment or inverted-persona models where harmful outputs occur alongside claims of alignment.
-
The Depth Ceiling: On the Limits of Large Language Models in Discovering Latent Planning
LLMs discover latent planning strategies up to five steps during training and execute them up to eight steps at test time, with larger models reaching seven under few-shot prompting, revealing a dissociation between discovery and execution.
-
LLM Evaluators Recognize and Favor Their Own Generations
LLMs show measurable self-recognition that linearly correlates with self-preference bias in evaluations, supported by fine-tuning experiments and controls for confounders.
-
Self-Recognition Finetuning can Prevent and Reverse Emergent Misalignment
Self-generated text recognition finetuning prevents and reverses emergent misalignment across multiple models by fortifying aligned character, unlike other finetuning baselines.
-
REFLEX: Self-Refining Explainable Fact-Checking via Verdict-Anchored Style Control
By steering activations along directions learned from base-vs-fine-tuned disagreements, REFLEX reports 64.99 macro-F1 on RAW-FC fact-checking with only ~465 self-refined contrastive samples and no retrieval.
-
Risk Reporting for Developers' Internal AI Model Use
A harmonized risk reporting standard for internal frontier AI model use, structured around autonomous misbehavior and insider threats using means, motive, and opportunity factors.