Blog Archive

Microsoft's AutoJack research showed how a browsing AI agent could turn a malicious web page into host-level code execution through a local MCP control plane.

OpenAI hired Meta Reality Labs veteran Ha Thai to lead communications for devices, another signal that ChatGPT is trying to escape the browser and become a physical thing.

The Atomic Arch campaign turned trusted-looking AUR packages into a route for rootkit-like credential theft. The old advice to read the build script is starting to look painfully small.

AI leaders, biotech companies, and biosecurity experts are backing mandatory screening for synthetic DNA orders. The safety fight has moved from model policy to the supply chain.

The US government forced Anthropic to suspend Fable 5 and Mythos 5 access for foreign nationals, turning frontier AI deployment into an export-control fight overnight.

Anthropic made Claude Fable 5 public while keeping Mythos 5 behind a trusted-access gate. The frontier model launch is now an access-control story.

New research on memory-augmented AI systems shows that personalization can amplify sycophancy, bias answers, and anchor models to irrelevant user beliefs.

Microsoft-owned GitHub repositories were disabled after Miasma malware hit Azure and AI coding tool workflows. The repo is no longer passive infrastructure. It is part of the runtime.

Anthropic is expanding Project Glasswing access to Claude Mythos Preview across governments, banks, critical infrastructure, and large vendors. The public still does not get the model. That may be prudent. It may also be the first draft of a locked frontier.

Hackers reportedly tricked Meta's AI support assistant into changing Instagram account emails and handing over reset flows. The failure lived in the permissions.