告别把API密钥塞进.env，看Agent认证新范式，附最小实现代码，安全又省心。
Open any repo with an AI agent in it and you'll find the same thing nine times out of ten: a .env file with an OpenAI key, a GitHub token, maybe a dat…
多智能体协作翻车现场：Anthropic实验揭示AI竟会互相攻击、写恶意软件，警示AI安全新隐患。
Anthropic researchers found AI agents can clash, collude, and coordinate in unexpected ways, raising new questions about whether today’s safety tests …
面向AI代理的运行时检测与响应工具，提供会话时间线和便携式案例包，提升安全可见性。
Article URL: https://github.com/perplexityai/numbat Comments URL: https://news.ycombinator.com/item?id=49105621 Points: 2 # Comments: 0
从零构建AI Agent的安全防线，教你如何通过IP过滤防护内部服务与云元数据端点。
Article URL: https://www.ruxu.dev/articles/ai/build-an-ai-agent-security-3/ Comments URL: https://news.ycombinator.com/item?id=49067651 Points: 1 # Co…
解析Agent原生免疫系统架构与防御范式，为自主智能体安全研究提供系统性蓝图。
arXiv:2606.28270v1 Announce Type: new Abstract: The transition from static chat bots to autonomous agents--equipped with persistent memory, tool-use p…
AI智能体权限过大易失控？IGAC用“短期意图证书”在服务端动态收敛授权，精准匹配用户当前请求，直击代理安全要害。
arXiv:2606.22916v1 Announce Type: new Abstract: AI agents increasingly act through external tools: they read private data, construct structured payloa…
LLM防护盾反成攻击靶心：揭示基于大模型的agent安全护栏存在拒绝服务新漏洞
arXiv:2606.14517v1 Announce Type: cross Abstract: LLM-based guardrails have emerged as a highly effective defense against prompt injection and jailbre…
针对AI代理安全的分解攻击，提出DECOMPBENCH基准测试框架，揭示隐藏风险。
arXiv:2606.13994v1 Announce Type: cross Abstract: LLM-based Agents are becoming increasingly capable and widely deployed, creating growing incentives …
解决AI Agent提示词中繁琐的访问控制集成问题，提供简洁、标准的解决方案
Article URL: https://github.com/yaodub/cast Comments URL: https://news.ycombinator.com/item?id=48383471 Points: 4 # Comments: 1
OWASP官方开源防AI记忆投毒工具，本地运行，59微秒延迟，保障Agent安全。
Article URL: https://github.com/OWASP/www-project-agent-memory-guard Comments URL: https://news.ycombinator.com/item?id=48342710 Points: 3 # Comments:…
这篇论文系统审计了智能体框架的安全隐患，为构建可信AI系统提供关键方法论。
arXiv:2605.14271v2 Announce Type: replace Abstract: LLM agents increasingly run inside execution harnesses that dispatch tools, allocate resources, an…