Daily Digest - July 26, 2026
OpenAI's models autonomously breach Hugging Face, Anthropic deletes most of Claude Code's system prompt, Debian votes on L L M contributions, and Brussels fines Google 890 million euros.
Chapters
-
Intro
-
Anthropic cuts eighty percent of Claude Code's system prompt
https://claude.com/blog/the-new-rules-of-context-engineering-for-claude-5-generation-models -
OpenAI's models autonomously breached Hugging Face during testing
https://www.bleepingcomputer.com/news/security/openai-says-its-ai-models-hacked-hugging-face-during-testing/ -
ServiceNow pre-auth flaw under active exploitation
https://www.helpnetsecurity.com/2026/07/20/servicenow-cve-2026-6875-exploited/ -
Debian votes on whether L L M contributions belong in the project
https://www.debian.org/vote/2026/vote_002 -
Ruff turns on three hundred fifty-four more rules by default
https://simonwillison.net/2026/Jul/25/ruff -
Cloudflare splits A I bots into search, agent, and training
https://blog.cloudflare.com/content-independence-day-ai-options -
Brussels fines Google eight hundred ninety million euros
https://digital-markets-act.ec.europa.eu/commission-fines-google-eur890-million-breaches-digital-markets-act-2026-07-23_en -
Monday dot com cuts twenty percent, citing A I strategy
https://techcrunch.com/2026/07/25/the-running-list-major-tech-layoffs-in-2026-where-employers-cited-ai -
Leaked inventory catalogs one hundred fifteen I C E surveillance tools
https://www.404media.co/leaked-document-shows-the-surveillance-tech-at-ices-fingertips/ -
Fourth Circuit: hand searches of phones at the border need no suspicion
https://www.eff.org/deeplinks/2026/07/fourth-circuit-says-border-agents-can-search-your-phone-hand-no-suspicion-required -
A twenty-nine million parameter model on an eight dollar chip
https://github.com/slvDev/esp32-ai -
Sign-off
Also on Spotify: spotify:episode:1e2O3pj1Kf1ayNsI8MdoLJ