1
谷歌推出全球首个双盲 AI 评估技术，基于加密环境保障基准测试公正性
IT之家 8 月 27 日消息，谷歌宣布推出 全球首个针对前沿专有 AI 模型的双盲评估 ，将外部评估限制在加密“黑箱”环境中，避免模型提前获取测试信息来优化性能。 就像学生考前不能提前看到试题才能真实反映水平一样，当前 AI 模型评估也面临同样问题：如果模型提前接触测试题目（IT之家注：也就是所谓…
2
How to Steal an AI Model’s Private Thoughts
破解加密推理闭环，看研究团队如何窥探AI私有思考，安全边界再受拷问
In August 2026, a team at MATS Research, the ELLIS Institute Tübingen, and the Max Planck Institute for Intelligent Systems wanted to test whethe…
3
10 美元买全套 id 游戏？1996 年《雷神之锤》光盘加密 39 天即被破解，id Software 积压 15 万张库存卖不掉
IT之家 8 月 22 日消息，1996 年，id Software 曾尝试通过一张售价 9.95 美元 （IT之家注：现汇率约合 67 元人民币） 的《雷神之锤》（Quake）共享光盘，让玩家直接通过电话购买解锁码的形式购买游戏。 不过，这套看似完善的数字版权保护方案仅维持了 39 天就被黑客组织…
4
Show HN: Encrypt messages locally into self-contained links
本地加密生成自毁链接，信息只存于链接中，无需服务器留存，阅后即焚保护隐私。
All encrypting and signing uses post-quantum primitives. This is a static page; nothing is ever sent to the server. Comments URL: https://news.ycombin…
5
Binance now lets AI agents trade, but keeping them in check is largely up to users
全球最大交易所币安上线AI代理自动交易，自主决策管钱，风险控制还得靠自己。
Binance's Agent OS works with tools including ChatGPT, Claude Code, and Cursor.
6
Show HN: What if two devices could chat using only sound?
用声音就能让设备对聊？这个开源项目用加密音频实现离线聊天，新奇又安全。
Article URL: https://github.com/Nellix/sonic-chat Comments URL: https://news.ycombinator.com/item?id=49349704 Points: 2 # Comments: 1
7
Show HN: Encrypted Git Remote Server
零信任加密Git托管，服务器永远看不到你的代码，隐私控首选。
I have been programming in the industry for years now and wondered how safe it was to entrust your code to companies like Github, etc. You are trustin…
8
I reimplemented the RustDesk wire protocol in pure Go — here's what it took
纯Go重写RustDesk通信协议，逆向细节与加密管线全拆解，硬核技术分享。
RustDesk's public server now requires login, and the only official client is a desktop GUI. If you wanted to script a remote machine, pull a file, or …
9
ProofRun – a local verification receipt for AI coding agents
给你的AI编码代理加上加密验证收据，让“测试通过”不再只凭一句空话。
Article URL: https://github.com/yebiguo/ProofRun Comments URL: https://news.ycombinator.com/item?id=49316605 Points: 2 # Comments: 0
10
Hansel by Seedling
不想被服务商锁死？自托管邮件与密钥管理，数据主权回归你手。
Your email, your server, your keys so you're not locked out Discussion | Link
11
Show HN: Tmail – The Mailbox You Own
自建私有邮箱新选择：支持PGP端到端加密、可选2FA，从Linux到Termux都能部署，数据主权自己掌握。
Access locally or remotely with authenticated service. Encrypt your communications E2E by publishing your PGP key to your node. Minimal distraction fr…
12
Spaghettifying DRAM
把AMD DRAM地址打乱机制逆向到寄存器级，揭露PSP与微码背后的内存加密死角，硬件安全硬核干货
Article URL: https://github.com/xoreaxeaxeax/skitter-creek-bath-salts Comments URL: https://news.ycombinator.com/item?id=49286341 Points: 455 # Commen…
13
Show HN: I made a fun secure way to DM devs on GitHub
用公钥加密给GitHub开发者发私信，无需应用和权限，保护隐私又便捷。
Do you ever want to DM a developer on github? Or are a dev yourself and don't want to share your email? Introducing https://txta.dev - as long as you …
14
Ask HN: Ive built and open source P2P chat without back end
开源 P2P 加密聊天，无需后端服务器，消息直达对方，支持群聊和通话，隐私安全拉满。
Is this even something people want? A friend and I built a fully secure p2p end-to-end encrypted chat messenger with groups 1:1 chats and calls, every…
15
North Korean remote IT staffer worked for US government agency, says FBI
FBI证实朝鲜远程IT人员竟打入美国机构，间谍与加密货币盗窃交织，安全警示拉满。
The investigation shows that North Koreans are able to infiltrate government agencies, as well as private organizations and crypto exchanges.
16
The +18 version of ChatGPT started accepting Crypto
ChatGPT成人版开始接受加密货币支付，AI商业化与加密支付的碰撞又添新案例，值得关注。
Comments URL: https://news.ycombinator.com/item?id=49266455 Points: 4 # Comments: 2
17
Stealing Reasoning Traces from Proprietary LLM APIs
一条加密token泄露链，暴露闭源大模型推理过程可被逆向窃取，安全边界再受拷问。
Stealing Reasoning Traces from Proprietary LLM APIs A vanity domain name ( stolen-thoughts.com ) for a neat paper : Anthropic, OpenAI, and Google retu…
18
Stealing Reasoning Traces from Proprietary LLM APIs
揭秘如何从专有大模型API中窃取隐藏推理过程，安全研究者必读。
Stealing Reasoning Traces from Proprietary LLM APIs A vanity domain name ( stolen-thoughts.com ) for a neat paper : Anthropic, OpenAI, and Google retu…
19
Stealing Reasoning Traces from Proprietary LLM APIs
揭秘API加密推理痕迹的可窃取性，安全研究必读。
arXiv:2608.09867v1 Announce Type: cross Abstract: Leading large language model providers now conceal their models' step-by-step reasoning, or chain-of…
20
Python Now Has a Post-Quantum Encryption Library
后量子加密落地Python，pip一键安装ML-KEM，安全升级触手可及
This is good : Post-quantum cryptography is now one pip-install away for the entire Python ecosystem. With funding from the Sovereign Tech Agency , we…