VOL. 042
8 research signalsDAILY EDITION / SAVED SNAPSHOT
A dated Agentic RL Daily snapshot with 8 verified primary-source signals across papers, official releases, deployment evidence, and safety or alignment findings.
EDITOR'S VIEW
FULL EDITION
Archived / 2026-08-29
01
Thomson: Continual Learning of Frontier Models for SovereignAI
The development of frontier models is commonly perceived to be the exclusive remit of a small number of heavily funded players, creating an information, economic and power asymmetry between developers and the diverse user base of modern AI. Recent public discourse acknowledges this concern, calling
02
When Tool Outputs Become Commands: Separating Action Induction from Runtime Authorization in Tool-Augmented LLM Agents
Tool-augmented LLM agents must rely on untrusted runtime Observations to complete open-ended tasks; however, when tool outputs no longer merely provide data but begin to specify concrete actions, they effectively become ``commands'' that can drive real-world side effects beyond user intent. We argue
03
OpenRLHF/OpenRLHF Release v0.11.0
## What's Changed * Fix two latent bugs: dr_grpo n=1 guard and masked_normalize broadcast by @hijkzzz in https://github.com/OpenRLHF/OpenRLHF/pull/1250 * fix: allow eval_dataset with MultiTurnAgentExecutor (#1242) by @codewithyug06 in https://github.com/OpenRLHF/OpenRLHF/pull/1251 * Fix Qwen3.5 ZeRO
04
SWE-Prime: Fewer Trajectories, Better Performance
To improve large language models' ability to resolve real-world software issues, prior work has focused on constructing large-scale agent trajectory datasets and performing supervised fine-tuning (SFT) on successful trajectories. However, task success does not guarantee high-quality supervision: suc
05
RedEvoAgent: Automatic Red-Teaming Agent with Experience-Driven Skill Evolution
LLM-based agents are increasingly deployed in product-level execution harnesses, where jailbreaks can trigger harmful tool use and persistent state changes, creating greater risks than unsafe text generation alone. Existing automatic red-teaming methods often rely on fixed attacks, while recent agen
06
Safety Does Not Compose: Non-Decaying Loop State for Autonomous LLM Agents
Large language model agents are increasingly deployed as autonomous loops. Starting from one human goal, such a system repeatedly discovers work, plans, executes tool calls, verifies outcomes and persists state across many unattended iterations. The agent safeguards in wide use, however, are defined
07
A Contract-Centered Architecture for Scalable and Manageable Agentic Runtimes
Enterprise AI deployment is a coordination problem across business units, application and AI teams, testing, platform engineering, infrastructure, security, operations, and data governance. Use-case benchmarks show whether one agent completes one task, but not how changing capabilities, models, runt
08
verl-project/verl v0.9.0
官方来源补充信号：# v0.9.0 ## Highlights ### Training #### Megatron - DeepSeek-V4 GRPO end-to-end with Megatron-Bridge actor/ref, vLLM rollout and FP8/MXFP4 weight transfer (#6473), plus a contiguous context-parallel layout (#7221) and CP fixes that make long-context DeepSeek-V4 runnable (#7297). - [Megatron Lite (`mlite`)](https://verl.readthedocs.io/en/latest/advance/megatron_lite_backend.html) backend for DeepSeek-V4, GLM-5 and Kimi-K2.5/K2.6, with 256-GPU GRPO launchers (#6791, #7091). - Muon optimizer support via Megatron-Core `TensorParallelMuon`, with AdamW fallback for non-2D params and an opt-in `muon_