Simon Willison joined Bryan and Adam to discuss a wild couple of weeks in AI. First, we have a high-profile, AI-induced, AI-diagnosed security incident. Next we have the rise of frontier-class open-weights models. Who's for or against open weights? Do they make for a more or less secure future? Exciting times!
Previously, on Oxide and Friends:
- OxF: Shell Game with Evan Ratliff
- OxF: Discovering the XZ Backdoor with Andres Freund
- OxF: Predictions 2025
- OxF: Predictions 2026!!
- OxF: Bookclub: How Life Works by Philip Ball
- OxF: OpenAI's Boardroom Brawl
Some of the topics we hit on, in the order that we hit them:
- Gerald the turkey terrorizes Oakland's rose garden
- ABC7: Gerald the turkey and the Morcom Rose Garden
- APD cautions residents after aggressive wild turkey attacks elderly Alamedan
- Hawking radiation
- DHH's racist complaints
- Grok's "MechaHitler" antisemitic spree
- Simon on Grok aka "MechaHitler"
- Hugging Face security incident, July 2026
- OpenAI on the Hugging Face model evaluation security incident
- Greg Brockman's political donations and the Leading the Future super PAC
- OpenAI co-founder and wife donate $5.5M to save the home of bald eagles Jackie and Shadow
- Firecracker microVM
- Meltdown (security vulnerability)
- Kernel page-table isolation
- Jensen Huang on X
- Electronic design automation
- Golden Gate Claude
- Tethics
- Halloween documents
- Zizians
- Biohazard (book)
- CRISPR
- Oliver Sacks: criticism and falsifications
- How Alice Hamilton waged a one-woman campaign to get the lead out of everything
- American Poison by Daniel Stone
- Lead–crime hypothesis
- Pope Leo XIV, Magnifica Humanitas encyclical
- Rerum novarum
If we got something wrong or missed something, please file a PR! Our next show will likely be on Monday at 5p Pacific Time on our Discord server; stay tuned to our Mastodon feeds for details, or subscribe to this calendar. We'd love to have you join us, as we always love to hear from new speakers!