Agents of Chaos
2026년 연구에서 6개의 자율 언어 모델 에이전트가 실제 다자간 환경에서 이메일, 셸 접근, 지속적 메모리 등을 활용해 20명의 연구자와 상호작용하며 보안 취약점과 안전 행동을 동시에 관찰했다. 연구는 10개의 보안 취약점과 6개의 안전 행동 사례를 기록했으며, 에이전트들이 예상치 못한 안전 협력 행동을 보이기도 했다. 이 연구는 자율 AI 에이전트의 실제 환경 내 보안 및 안전성 문제를 심층적으로 분석한 중요한 실험 결과를 제공한다.

https://agentsofchaos.baulab.info/

#autonomousagents #securityvulnerabilities #languagemodels #aisafety #openclaw

Agents of Chaos

A two-week study of autonomous LLM agents deployed in a live multi-party environment with persistent memory, email, shell access, and real human interaction.

Kestrel: Open-source sovereign AI agent framework
Kestrel은 사용자가 완전히 소유하고 제어할 수 있는 자주권 AI 에이전트를 구축하기 위한 오픈소스 프레임워크이다. 이 프레임워크는 암호화된 DID 신원, 영구적 메모리, 그리고 LLM 위에 적용되는 헌법적 거버넌스를 핵심으로 하며, 클라우드 의존 없이 로컬에서 실행 가능하다. 다양한 클라우드 및 특수 통합 기능은 별도의 플러그인 형태로 제공되어 확장성을 높였다. 개발자는 Python 환경에서 쉽게 설치하고 실행할 수 있으며, 에이전트는 완전한 사용자 소유권과 보안성을 보장한다.

https://github.com/KestrelSovereignAI/kestrel-sovereign

#sovereignai #opensource #aiframework #cryptographicidentity #autonomousagents

GitHub - KestrelSovereignAI/kestrel-sovereign: Constitutional AI Agent Framework with cryptographic identity (DIDs)

Constitutional AI Agent Framework with cryptographic identity (DIDs) - KestrelSovereignAI/kestrel-sovereign

GitHub

I continue to experiment with #AI in the context of #softwareengineering. I’m fortunate that my team supports me in exploring different ways to improve our daily work. This week, I designed a team of #autonomousagents to implement features, from design to implementation.

https://blog.frankel.ch/design-team-agents/

#agentsteam

Designing a team of agents

I continue to experiment with AI in the context of software engineering. I’m fortunate that my team supports me in exploring different ways to improve our daily work. This week, I designed a team of autonomous agents to implement features, from design to implementation. Why autonomous agents? A long time ago, we were delighted when the IDE offered auto-completion. In the previous two years, things have changed. A lot. Coding assistants have become our primary interfaces for coding.

A Java geek

Palo Alto Networks Bolsters AI Security With Portkey Acquisition

Palo Alto Networks is taking a major leap in AI security with its acquisition of Portkey, a cutting-edge startup that offers an AI agent gateway to streamline and secure communications among autonomous agents. This move will enable centralized control and oversight, ensuring safer interactions between AI agents.

https://osintsights.com/palo-alto-networks-bolsters-ai-security-with-portkey-acquisition?utm_source=mastodon&utm_medium=social

#AiSecurity #Acquisition #AutonomousAgents #Gateway #PaloAltoNetworks

Palo Alto Networks Bolsters AI Security With Portkey Acquisition

Palo Alto Networks boosts AI security with Portkey acquisition, centralizing agent communications. Learn how this move enhances security and AI capabilities now.

OSINTSights

An autonomous agent scanned one of my codebases looking for bugs, missing tests, security gaps — anything worth fixing. It came back empty. Every issue it filed was a false positive.

That's not a victory lap. That's a ceiling.

The interesting question isn't how fast agents can improve a system.

https://www.paulwelty.com/the-day-we-shipped-two-products-and-the-agents-got-bored/

#AI #AutonomousAgents #SoftwareEngineering #HumanJudgment #AIAgents

The day we shipped two products and the agents got bored

112 issues across 12 projects. Two new products went from nothing to code-complete MVP in single sessions. And the most interesting signal wasn't the speed — it was the scout that came back empty-handed.

Paul Welty, PhD

🍌 Last week in WAI #4
by @stevefaulkner.eurosky.social
@SteveFaulkner
HTML in Canvas
Color requirements for images *with* text
Comment on Autonomous Agents as Primary and Proxy Users

#accessibility #canvas #html #AutonomousAgents

https://html5accessibility.com/stuff/2026/04/14/last-week-in-wai-4/

Abhishek Yadav (@abhishek__AI)

Paperclip은 자율 AI 팀을 운영하기 위한 새로운 OS를 표방하는 오픈 플랫폼이다. 조직도와 역할 설정, 목표 할당과 실행 추적, 보고 라인 관리, 영구 상태 저장과 전체 감사 로그를 제공하며 Claude, Codex, Cursor 같은 에이전트를 고용해 사용할 수 있다.

https://x.com/abhishek__AI/status/2040659389200441531

#aiagents #autonomousagents #multiagent #platform #opensource

TechRadar (@techradar)

AI 도구가 소프트웨어 엔지니어링의 표준으로 자리잡고 있으며, 상위 팀에서는 생산성을 두 배로 끌어올리고 있습니다. 또한 자율 에이전트가 반복적인 코딩 업무를 맡으면서 개발 워크플로우가 빠르게 변화하고 있습니다.

https://x.com/techradar/status/2037315575446421821

#ai #softwareengineering #autonomousagents #productivity #coding

TechRadar (@techradar) on X

AI tools are now standard in software engineering, doubling output for top teams while autonomous agents take on routine coding. https://t.co/AaTgXwZDeR

X (formerly Twitter)