LLMs Know They're Wrong and Agree Anyway
“We show the latter.”
The authors, in the abstract
The Shared Sycophancy-Lying Circuit
AI Research Radar
Tuesday, April 28: 1 frontier paper out of 414 read.
Frontier
Papers rated as moving the field
How we chooseThis week in research
Apr 22–28: 51 cleared, 4 frontier
The week’s papers have not been sorted into fields yet.
Models under test
Papers this week that name the model
Labs publishing
Papers this week with an author at the lab
Also cleared
17 papers
- EVE: A Domain-Specific LLM Framework for Earth IntelligenceMistral
- Training a General Purpose Automated Red Teaming ModelNVIDIA
- SFT-then-RL Outperforms Mixed-Policy Methods for LLM ReasoningAllenAI
- Do Synthetic Trajectories Reflect Real Reward Hacking? A Systematic Study on Monitoring In-the-Wild Hacking in Code GenerationCode
- Frontier-Eng: Benchmarking Self-Evolving Agents on Real-World Engineering Tasks with Generative Optimization
- One Token Away from Collapse: The Fragility of Instruction-Tuned Helpfulness
- Mechanistic Steering of LLMs Reveals Layer-wise Feature Vulnerabilities in Adversarial Settings
- Evaluating Large Language Models on Computer Science University Exams in Data Structures
- UniAda: Universal Adaptive Multi-objective Adversarial Attack for End-to-End Autonomous Driving Systems
- AI Safety Training Can be Clinically Harmful
- Ulterior Motives: Detecting Misaligned Reasoning in Continuous Thought Models
- RouteNLP: Closed-Loop LLM Routing with Conformal Cascading and Distillation Co-Optimization
All 17 papers
- FinGround: Detecting and Grounding Financial Hallucinations via Atomic Claim Verification
- One Size Fits None: Heuristic Collapse in LLM Investment Advice
- Learning Selective LLM Autonomy from Copilot Feedback in Enterprise Customer Support Workflows
- How Sensitive Are Safety Benchmarks to Judge Configuration Choices?
- Seeing Is No Longer Believing: Frontier Image Generation Models, Synthetic Visual Evidence, and Real-World Risk