AI Research Radar
Wednesday, June 10: no frontier papers out of 364 read.
Frontier
Papers rated as moving the field
How we chooseNone of the 38 cleared papers that day was rated frontier.
This week in research
Jun 4–10: 152 cleared, 7 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
38 papers
- Local Is Not a Sufficient Privacy Boundary: Governing OS-Integrated On-Device AIGoogle
- Benchmarking and Exploring the Capabilities of LLMs for Attack InvestigationsGoogle
- CIAware-Bench: Benchmarking Control Intervention Awareness Across Frontier LLMsGoogle
- Gradient-Guided Reward Optimization for Inference-time AlignmentCode
- RKSC: Reasoning-Aware KV Cache Sharing and Confident Early Exit for Multi-Step LLM InferenceCode
- SpeechJBB: Probing Safety Alignment and Comprehension in Large Audio Language Models under Code-Switched Speech
- Durable Evaluation Framework: Adversarial Arbitration for Sycophancy Reduction in Large Language Models
- The Interlocutor Effect: Why LLMs Leak More Personal Data to Agents Than Humans
- From Confident Closing to Silent Failure: Characterizing False Success in LLM Agents
- Alignment Collapse Under KV Cache Quantization: Diagnosis and Mitigation
- PreAct-Bench: Benchmarking Predictive Monitoring in LLMs
- Bypassing Copyright Protection in Diffusion-based Customization via Two-Stage Latent Feature Optimization
All 38 papers
- GitInject: Real-World Prompt Injection Attacks in AI-Powered CI/CD Pipelines
- A Controlled Audit of Pretraining Contamination in Public Medical Vision-Language Benchmarks
- VFUSE: Virulent Feature Understanding with Sparse autoEncoders
- Gaming AI-Assisted Peer Reviews Poses New Risks to the Scientific Community
- Flow Control: Steering Vision-Language-Action Models with Simple Real-Time Inputs
- Alignment Defends LLMs from Property Inference Attacks
- SHAPO: Sharpness-Aware Policy Optimization for Safe Exploration
- OpenRTLSet: A Fully Open-Source Dataset for Large Language Model-based Verilog Module Design
- MIRAGE: A Polarity-Flipping Encoding Subspace in LLM Agents
- Catching One in Five: LLM-as-Judge Blind Spots in Production Multi-Turn Transaction Agents
- Test-time Adversarial Takeover: A Real-time Hijacking Interface against Robotic Diffusion Policies
- UMI-Bench 1.0: An Open and Reproducible Real-World Benchmark for Tabletop Robotic Manipulation with UMI Data
- SpenseGPT: Practical One-shot Pruning Enabling Sparse and Dense GEMMs for LLM Inference
- Stop Early, Spend Less: Hidden-State Probes as a Practical Recipe for Streaming Moderation of LLM Outputs
- Uncovering Vulnerability of Vision-Language-Action Models under Joint-Level Physical Faults
- Assessing Automated Prompt Injection Attacks in Agentic Environments
- Speaker Group Encoding in Self-supervised Speech Recognition Models
- Accounting for AI Inference in Corporate GHG Inventories: A Four-Tier Methodology for Scope 3 Category 1 Reporting
- MemVenom: Triggered Poisoning of Multimodal Memories in Web Agents
- Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluation
- On-sky demonstration of reinforcement learning for adaptive optics control
- Ethical and Technical Limits of Deepfake Speech Datasets
- Democratising Camera Trap AI: An Open-Source Model for Detecting UK Mammals
- The Shibboleth Effect: Auditing the Cross-Lingual Distributional Skew of Large Language Models
- PhantomBench: Benchmarking the Non-existential Threat of Language Models
- ABC-Bench: An Agentic Bio-Capabilities Benchmark for Biosecurity