一套混淆管线搞定全平台应用加固，云构建服务内直接对合并JAR处理，安全升级就看这波。
Obfuscating only the Android artifact is a poor security model for a cross-platform application. Codename One App Hardening transforms the merged appl…
用对抗性代码混淆技术防御大语言模型分析，Acoda方法为LLM安全带来全新视角
Article URL: https://arxiv.org/abs/2606.11755 Comments URL: https://news.ycombinator.com/item?id=49109869 Points: 2 # Comments: 1
苹果与OpenAI诉讼前关键细节曝光，两封邮件间隔13分钟与一次姓名混淆暴露沟通僵局。
IT之家 7 月 16 日消息，NBC 昨日（7 月 15 日）发布博文，在苹果和 OpenAI 公司的商业纠纷案件中， 2 封关键邮件、13 分钟间隔与 1 次姓名混淆，构成双方起诉前沟通受阻的核心细节。 IT之家本月报道， 苹果公司上周在美国加州北区联邦法院起诉 OpenAI ，指控该公司数名员…
用 Shadow DOM 实现内容混淆而不牺牲用户体验，优雅兼顾防爬取与前端性能，代码值得一看。
Hey HN! After reading this article about email obfuscation [0] a few weeks ago, I build html-obfuscator. And I'm awfully proud of the result! I know o…
PromptPET通过智能混淆保护提示词隐私，在不明显牺牲回答质量的前提下平衡安全与效用，适合处理敏感查询。
arXiv:2607.02932v1 Announce Type: cross Abstract: Privacy is an important challenge when users interact with AI chatbots, since users may share sensit…
LLM agent工具调用暗藏越权风险，深度审计主流框架的“能力门”与授权混淆问题，开发者必读。
arXiv:2606.28679v1 Announce Type: cross Abstract: Tool-using LLM agents increasingly read untrusted content while holding side-effecting tools such as…
用位平面混淆技术保护图像隐私，兼顾可用性与安全性，视觉AI隐私保护新思路
arXiv:2606.29417v1 Announce Type: new Abstract: The unprecedented growth of computer vision applications, such as surveillance systems and social medi…
新方法GELO专为共享加速器场景设计，有效保护开源大模型的提示词隐私，对抗内存读取攻击。
arXiv:2603.05035v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) are increasingly served on shared accelerators where an adversa…
对比研究 GPT-3.5 与 GPT-4 如何利用大模型混淆代码风格，为匿名代码与隐私保护提供新思路。
arXiv:2606.22306v1 Announce Type: cross Abstract: In the rapidly evolving field of software development, code stylometry analyzing unique stylistic si…
谷歌Gemini语音助理曝出新型注入漏洞，黑客通过特殊通知隐藏恶意指令，可远程操控AI执行非授权操作。
IT之家 6 月 7 日消息，安全公司 SafeBreach 披露，谷歌 Gemini 存在一种新型安全漏洞。黑客可通过 WhatsApp、短信等渠道发送特殊构造的通知信息，并将恶意指令隐藏在其他语言文本或“静音超链接”（muted hyperlinks）中，从而诱导 Gemini 执行未经授权的操…
A*算法驱动多智能体，对LLM提示发起常识混淆攻击，揭示大模型语言理解的脆弱性。
arXiv:2606.01441v1 Announce Type: new Abstract: Large language models (LLMs) excel in reasoning and knowledge-intensive tasks but remain vulnerable to…
Python代码易暴露，用混淆技巧为算法与商业逻辑穿上“铠甲”
You've built something valuable in Python. Maybe it's a proprietary algorithm, a business logic engine, or an AI application with carefully crafted pr…
探讨AI代理陷入权限困境的核心矛盾，以及用LLM监控的实践思路
An agent's value is proportional to the permissions it's been granted. There's been a lot of hype around solutions like default denial proxies, key va…
解读猫语的AI模型，用多模态大模型破解猫的语义混叠之谜，动物行为学研究新突破
Article URL: https://arxiv.org/abs/2605.09152 Comments URL: https://news.ycombinator.com/item?id=48234894 Points: 2 # Comments: 0