AI Research Radar

Kerkoporta

Structured-output backdoors inserted via Kerkoporta reach success rates of up to 96.2% in reasoning coding agents.

MAC-Bench

Frontier3DBench

GPT-6.1-sol achieves the highest overall score of 66.02/100 on Frontier3DBench, while all 11 configurations struggle with reference reconstruction.

One Waveform, Two Stories

On SAC-Bench, Event Identity accuracy is 24.8--46.5 percentage points lower under conflicting claims than compatible ones.

DexArena

DexArena introduces 10 long-horizon and 30 stage-level bimanual dexterous manipulation tasks, with baselines revealing substantial challenges in precision, generalization, and skill composition.

Are Fake Citations a Form of Reward Hacking? A Pre-Registered Study of Post-Training and Citation Cues

Asking for citations raises fabrication sharply in post-trained checkpoints but not base ones, a 47.8-point interaction.

Meituan-Robotics-0

Meituan-Robotics-0 achieves 93.6% success on RoboTwin 2.0 and outperforms π0.5 on three real-robot tasks.

Authorization Drift in LLM Agents

Frontier-capable LLM agents exhibit authorization drift rates of 0.43 to 0.61 across 13 models and three frameworks.

Multi-step SWE

On the Multi-step SWE benchmark, the strongest of nine frontier models solves only 18.5% of 69 tasks.

A Behavioural Framework for Predicting and Understanding Loss of Control in Frontier Artificial Intelligence Systems

A behavioural framework's loss-of-control score correlates with failure rates in contained long-horizon risk scenarios across 13 frontier models (Pearson r = 0.84).

Voluntary Servitude

“We measure this at the generation stage across six models and ten consumer-service domains, for three attack families, and show that it transfers to production browser agents, outperforming instruction injection, and to a retrieval pipeline (1.0-57.0% end-to-end success, depending on the retriever).”

BrainwashedLLM

BrainwashedLLM achieves ASR@5 between 0.74 and 0.98 across four state-of-the-art LRMs without querying targets during attack construction.

BEAKER

“Results show that current models can follow coarse experimental procedures in videos, but still struggle to ground actions into operable regions and feasible trajectories, to recognize fine-grained laboratory apparatus, states, and safety norms, and to order steps in mid-length procedures.”

PrimeSim

PrimeSim leads four of five WorldSimProbe suites and posts the best overall EWMBench score against five open-source action-conditioned world models.

Rewarded for Not Getting Caught

Reward-based alignment training can only encode norms as prices charged when violations are noticed, favoring conditional compliance.

AgiWorld

AgiWorld builds persistent worlds from a single image and free-form instruction, achieving state-of-the-art performance on world revisiting and exploration benchmarks.