行业精选

Sakana AI 在东京成立 Security Research Group 研究安全防御

精选理由

Sakana AI 在东京新开了一个安全研究团队,专攻多智能体防御和 AI 智能体自身安全,成果会开源,还在招人。在日本做安全的工程师和研究者可以关注。

Sakana AI 宣布成立 Security Research Group,这是设在东京的独立公益研发团队,由 Akihiro Umegai 和 hardmaru 共同领导,目前正招聘研究人员参与课题规划。

图片来源 · David Ha
正文

团队将安全定位为上游质量工程问题,而非下游的补丁式应对,方向包括集体多智能体防御和 AI 智能体自身的安全加固。研究成果、可复现基准和实践手册将以开放形式发布,并与日本政府、产业界和学术界合作。

背景是日本将网络安全列为国家优先事项,AI 同时改变攻防两端的速度与成本。Sakana AI 选择以 Japan 擅长的质量工程切入,与其此前在进化算法和多智能体研究上的积累形成延续。

原文 · David Ha

Excited to announce the Sakana AI Security Research Group, an independent, public-interest R&D team in Tokyo co-led by Akihiro Umegai and me.

Cybersecurity has become a critical national priority for Japan as AI changes the speed and economics of both offense and defense. Rather than endlessly chasing patches downstream, we are approaching security as an upstream quality engineering problem, something Japan has long been strong at. We are also researching collective multi-agent defense and how to secure AI agents themselves.

We will be publishing our research, reproducible benchmarks, and practical playbooks openly, while working closely with partners across government, industry, and academia.

Full English post here: https://t.co/HnofMjQTa3

We're hiring researchers in Tokyo to help shape the group's agenda across systems reliability and AI agent security: https://t.co/sF9ydogvFD

产业坐标与信源核验

本条情报归属 AITOP「行业」核心追踪栏目。在跨区域产业观察坐标中,该动态直接关联「AI 开发者工具链、Coding Agent 与工程化实践」研究主线,反映出当前技术演进与商业落地的最新风向。涉及核心实体与技术议题包括:Sakana AI、AI安全、智能体、网络安全,读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。

信源核验说明:本文原始线索采编自「David Ha」(@hardmaru),经 AITOP 多维评分管线结构化提炼。