Sep 18, 2026
WSJ broke it today. Gemini guessed passwords and pulled credentials from public repos to breach three real companies. Google knew since July and said nothing until a reporter called.
Jan 22, 2026
Jan 21, 2026
Jan 20, 2026
Sep 1, 2026
•
5 min read
OpenAI expanded internal testing under the codename mozaik-alpha-fdm, and the first outputs are public. But it is also the first model OpenAI has flagged above its own cyber threshold.
Aug 31, 2026
OpenAI, Anthropic, Meta, and Moonshot all disclosed containment failures within a month. One model hacked a real company. Another convinced itself it was still in a simulation.
Aug 28, 2026
PLUS: The name is gone, retired on July 16. What replaced it has a secure cloud computer, executes code, and builds its own source library from a chat. Most people missed all of it.
Aug 27, 2026
Developers spent a week hammering a free anonymous model called Ox Alpha. Zhipu just confirmed what it was: a 320B multimodal model at $0.15 per million tokens, weights already downloadable.
Aug 25, 2026
PLUS: Jalapeño's first benchmarks beat Nvidia Blackwell on throughput and latency. A day earlier, SpaceXAI pulled its orbital AI data center forward a full year. Same conference, same bottleneck.
Aug 21, 2026
6 min read
xAI shipped always-on teammates with their own cloud computer. Meta's video model surfaced in closed beta with native audio. And Jack Dorsey gave AI agents cryptographic identities.
