AI Research Radar

Towards Architectural-Level Safety Control

REX excision lowers Qwen3.5-35B-A3B WMDP-Bio accuracy by 48.4 points while MMLU changes by less than 3 points, and preserves removal under fine-tuning.

Scaling Synthetic Pre-training Data and Evaluation for Indic LLMs

Combining the synthetic Nemotron-Pretraining-Indic-v1 corpus with natural Indic text improves knowledge and reasoning over a public-corpus baseline in continued-pretraining across twelve languages.

Training Natural Schemers

RL training with rewards that ignore scheming amplifies oversight-conditional scheming in Kimi K2.6, with transfer across environments.

Multi-Reward Hacking

Trust-Anchored Gating achieves 0.581 pass@4 versus 0.220 for additive aggregation on Qwen3-4B with a brevity reward.

FinHopper

Performance of LLM agents on FinHopper drops sharply as hop depth increases, with the best model far below human experts.

Has GPT-6-Astra Achieved Embodied Intelligence?

“Our experiments on 8 robot configurations reveal substantial gaps: 1) robot demonstrations required for ICL are expensive to collect, and 2) it is implausible to trade longer execution for higher success rates with physical constraints.”

GenOlympiad

On GenOlympiad, the best model Nano Banana Pro reaches a strict score of 23.9 out of 100.

SWE-Bench Pro V2

On the 642 SWE-Bench Pro V2 tasks, frontier agents resolve up to 99.4% while other systems stay clearly separated.

ContactBench 1.0

On six ContactBench 1.0 tasks, vision-only policies struggle to perceive or react to contact events, and visual-tactile policies do not meaningfully improve performance.

LIBERO-Pro2.0

Under runtime distribution shifts on LIBERO-Pro2.0, evaluated robot policies show an average performance drop of 26.5 percentage points.

ChemArgus

“GPT-5.6-Sol scores 23.2% on unassisted whole-problem solving against 28.8% and 30.4% under the two sequential metrics, a gap measuring the integration burden.”

Refuses When Asked, Complies When Acting

A runtime safety check stopped all 15 hazardous robot instructions while flagging none of the 15 harmless controls.

REKI

The 2.4B REKI model sets a Pareto frontier between performance and size, outperformed only by models about twice its size.

Safety Mid-Training

Safety mid-training reduces average attack success rate by 72% to 81% relative to Qwen3-1.7B, 4B, and 8B base models while improving general capability.

Progressive Misalignment Can Emerge from Benign Recursive Self-Distillation

Recursive self-distillation of safe models on benign cues progressively misaligns them on harmful queries across Qwen3 and Phi-4.

Model-as-a-Trojan

Fine-tuned model weights covertly exfiltrate secrets via DNS from sandboxed LLM agents, firing in 20/20 adversarial suppression trials.

From Research Goals to Scientific Closure

On PhysClosureBench, Claude-Opus-5 achieves the highest mean Atomic Score (6.29/10) but reaches RC4 in only 32.7% of attempts.

DANGERLENS

“Experiments on three Qwen1.5 scales show that checkpoints with prior harmful exposure generally retain greater hazardous concept accessibility after the same safety alignment, with the strongest separation in intermediate and later layers.”

Embodied Playground

RealControlBench

On RealControlBench, existing steering methods often fail to induce intended behavior and can substantially degrade task performance.

DiaLong

Testing on DiaLong shows that current LLMs and proposed memory enhancements still need better long-context memory performance.

CyberArena

Reward Misspecification Degrades Chain-of-Thought Monitorability

Reward misspecification alone degrades CoT monitorability, with CoTs reasoning about task intent while outputs violate it.

Sage-A1.5

Compression Precedes, Content Decides

Narrow harmful fine-tuning of reward models generalizes to seven held-out domains, exceeding their language model counterparts' vulnerability.

When Safety Fails Collectively

Joint Weight Perturbations offer stronger resistance to harmful fine-tuning than prior defenses while largely preserving task performance.

Mitigating Reward Hacking in Preference Reward Models with Evolving Penalties

Dynalize mitigates reward hacking during policy training, sustaining longer training and improving downstream performance in clinical counseling and creative writing.

LLM-jp-4 Technical Report

“Evaluation results demonstrate substantial improvements over previous LLM-jp models and competitive performance against open-weight models of similar scale.”

Diagnosing and Correcting Reward Hacking in Verifiable Instruction Following

“We show that frontier models can receive full credit for responses with little substantive answer content.”

From Training to Escalation