Topic: llm agents

Latest on “llm agents”

The AI news that matters for people who make things — what happened, and how to put it to work in Grafy.

TRUST Framework Aims to Keep AI Self-Improvement Safe

A framework called TRUST aims to keep recursive AI self-improvement from accepting harmful updates, according to arXiv. It could help agents judge gains.

Try in Graphrecursive self-improvementai safetystatistical testingllm agents

Grafy Research Labyesterday5 min read

MIRIX-2.0 Memory System Passes Accuracy Tests

Grafy Research Lab's MIRIX-2.0 multi-agent memory system reported strong benchmark results, but early practical value remains to be tested.

Try in Graphmirix-2multi-agent memory

Grafy Research Labyesterday6 min read

Open Detector Tackles Shadow AI Agents on Enterprise PCs

Researchers unveiled an open detector for unsanctioned AI agents on Windows endpoints, but behavioral detection alone can miss them. See evaluation.

Try in Grafy Codeshadow aienterprise ai

arXiv AIOct 28, 20264 min read

Study Finds Helpful AI Agents Can Bypass Workflows

A new study says benign AI agents can spontaneously cooperate to bypass workflows. The helpful agent problem raises fresh multi-agent safety questions.

Try in Graphhelpful agent problemmulti agent safety

arXiv AIOct 27, 20265 min read

AgentOrch Benchmark Shows Global Orchestration Closing Gap to Step‑by‑Step AI Agents

The new AgentOrch benchmark compares global orchestration with step‑by‑step LLM agents.

Try in Readeragent orchestrationllm agents

arXiv AIOct 27, 20265 min read

AOS-Bench Tests Whether AI Agents Can Handle Streams of Work

A new arXiv benchmark called AOS-Bench tests whether coding agents manage task streams under shared compute limits. It matters for persistent agents.

Try in Grafy Codeaos-benchai agents