Tim Culpan reports TSMC may help run Terafab’s planned Texas chip factories; Musk calls it “just discussions.”
TechCrunch reports David Robinson led safety-report writing for major OpenAI launches and urged AI labs to learn from aviation and nuclear safety.
On an a16z podcast, Replit’s Amjad Masad describes using one company-wide agent; OpenRouter’s Alex Atallah argues for specialization.
YouTubeAxios quotes Bessent calling “industrial distillation” a nice word for stealing and claiming Kimi says it is Claude.
The Information reports coding agents account for about half of Vercel’s new business, up from less than 3% at the start of 2026.
Its creator says the app shows what their coding agents did.
Aleph Alpha says Kolibri activates about 3.5 billion of its 78 billion parameters per token.
AMD mentioned the commits alongside coffee and news, then cheered, “Open source FTW.”
The author says “no moat anymore” is why they build many small apps and links to a video.
The Information reports the companies are grappling with who pays when an agent makes the wrong purchase.
XDA Developers says changing both settings can improve the DeX experience, particularly for Fold 8 users.
9to5Google reports that free Gemini users will be limited to Flash-Lite starting October 9, 2026.
Engadget suggests turning one into a smart-home control panel or a companion for outdoor adventures.
Engadget says the 3DS has a lot to offer, especially if you haven’t experienced its game library.
PCMag says Apple Intelligence is on by default in iOS, but you can turn it off entirely with a quick tap.
Reuters reports that three sources said Jordanian authorities detained Saif al-Din Khader, whose alleged hacker alias is “Rey.”
Linuxiac reports that contributors must confirm their COSMIC pull requests contain no LLM-generated code, comments or descriptions.
Engadget recommends customizing Camera Control for more precision and creative control than Apple’s defaults allow.
Techmeme, linking to TechCrunch, says the plans could make ChatGPT a place to discover, launch and use software.
@sama says ascribing religious force to AI models or surrendering human judgment to them is a real safety issue.
