Clara Vásquez

A multi-day analysis of Hacker News’ top 60 stories reveals how its ranking really works: a public formula based on votes and time, plus a hidden penalty system tied to words, domains, and controversy. Here’s the formula, the code to reproduce it, and the rules no one officially documents.

A team of researchers chained a libheif vulnerability in OpenAI’s Discourse forum with an SSO flaw to take over employees’ ChatGPT and Codex accounts. In under 72 hours, they even opened a pull request inside the company’s internal repository.

A developer combined a $20 4G hotspot, a Clicks keyboard, and a SHARP display to build a Linux messaging device using the OpenStick project. Here’s how it went, from EDL mode to the custom adapter board.

A cohttp maintainer published a security patch and within ten minutes was already receiving automated probes: this is bugonomics, the phenomenon where a mere bug rumor is enough to generate an AI-built exploit. The average exploitation time is now -7 days, before the public patch is even available.

The DGFiP, France’s tax authority, confirmed a leak of 678,000 taxpayer records following a compromised VPN account. Access was cut off in late June, but the exfiltration was only detected when the data appeared for sale in August.

Cryptographer Matthew Green warns that AI specialized in hunting vulnerabilities is patching software so fast that police and intelligence agencies could lose their ability to hack encrypted devices. The phenomenon revives the ‘going dark’ debate the FBI opened in 2014.

A researcher documented a hardware backdoor active by default in VIA C3 x86 processors, capable of letting unprivileged code rewrite the kernel. The Rosenbridge project, published on GitHub, includes the tools to detect and close it.