资讯与观察· 社区热议
AI 行业动态、工程实践、产品方法和社区故事。
Dynamic Abliteration: Non-Destructive Refusal Suppression via Engram Steering
When working with open-weight LLMs like Qwen, controlling refusal behavior on security, administrative, prompts typically requires fine-tuning or permanent weight update. Traditional weight abliteration technique neutra…
Unknown number of Texas voter registrations went unprocessed due to DPS error
For more information, please read Texas counties are receiving tens of thousands of unprocessed voter registration applications due to state error.
12 分钟 · Hacker News
Forging 1024-bit RSA signatures in nearly SNFS time [pdf]
https://arstechnica.com/security/2026/09/theres-a-new-way-to...
1 分钟 · Hacker NewsLATEST STORIES
最新内容
Best LLM for every budget, updated daily精选导读
暂无摘要
Google’s Project Suncatcher to put ML infrastructure in space精选导读
https://www.nytimes.com/2026/09/24/technology/google-suncatc... (https://archive.ph/BGFnP)
Is A.I. Above the Law?精选导读
What happened—which can be reconstructed from messages that the agents sent one another and from traces of their own internal reasoning, known as chain-of-thought records—makes for gripping reading. The story has a nice…
Jev Is Not a Language Model, but It Breaks Like One精选导读
A new kind of AI model returns decisions instead of text. We spent a day trying to change its mind. It cost about 50 cents.
Oracle Cites 'Force Majeure' to Shield Itself on Controversial Data Center精选导读
暂无摘要
Owners mourn spoiled food after firmware update bricks Samsung smart fridges精选导读
暂无摘要
'That's so AI ' What gen Alpha's biggest insult tells us精选导读
‘OMG my mum and dad are so AI.’ Photograph: Posed by models; Iuliia Bondar/Getty Images
What Is RLCD? The Secret Behind Jev精选导读
From pairwise reward modeling to calibrated, multiway decisions
Hackers influence ChatGPT and Gemini to direct users to scam centers精选导读
Dark Sourcery: How Hackers Manipulate AI to Scam You