OpenAI leads today with two posts: the agreement to acquire Pierre Computer Company, the small team whose Git trees, diffs, and Code Storage service already power parts of Codex, and a write-up of its autonomous Linux-kernel vulnerability pipeline, which validated 948 local privilege-escalation exploits for about $1.2 million in API costs. Beyond AI, three patch-now security stories: a man-in-the-middle flaw in APT that lets a malicious mirror swap verified Debian and Ubuntu packages and run code as root, Xen hypervisor bugs — found using Anthropic’s Claude — that let one guest VM read and write another’s memory, and Binance’s post-mortem of the $6.43 billion theft that began with fake job interviews and an abused internal AI tool.
From the labs
- OpenAI to acquire Pierre Computer Company, Inc. — OpenAI
- A thousand fewer paths to root — OpenAI (autonomous Linux vulnerability discovery pipeline)
Beyond AI
- Debian and Ubuntu APT Flaw Lets MitM Attackers Swap Verified Packages and Run Code as Root — The Hacker News
- Xen Hypervisor Flaws Let One Malicious VM Read and Write Any Other VM’s Memory — The Hacker News
- Binance Reveals How Fake Job Interviews and AI Tool Abuse Led to $6.43 Billion Theft — The Hacker News