What you need to know
5 storiesThe short versionAnthropic released FlashProve so anyone can check which model answered a prompt, and Moonshot put the weights of its 2.8 trillion parameter Kimi K3.1 Pro on Hugging Face.Full rundown
Right now you can't tell which model actually answered your prompt. Anthropic's Frontier Red Team wants to fix that with FlashProve, open-source code it put out on November 2 that attaches a proof to each reply. The proof is 19 MB and checks in seconds without a GPU. That's enough to catch a router quietly swapping in a cheaper model.
Kimi K3.1 Pro is out. Moonshot put its weights on Hugging Face on November 4, so you can run the 2.8 trillion parameter model yourself (if you're one of the biggest model hosts, though, you'll need a separate deal first).
Apple opened Siri AI to any developer on November 3, through App Intents in the iOS 27.3 betas. Nobody can use it yet. App Store submissions open in the spring.
Google will switch Gemini and Search off for K-12 students until a school administrator turns them back on. Britain's cyber agency thinks AI-powered malware is highly likely to be used in a ransomware attack on the country's critical infrastructure by 2028. And Texas elected Gina Hinojosa, who wants a pause on new data centers. Elon Musk, who'd given her opponent $5.5 million, then posted that space was looking appealing for data centers.
Update Chrome. The new stable release fixes seven critical memory flaws, and a crafted web page could use some of them to get outside the sandbox. Got an old Mac or iPhone that skipped Apple's May 2024 updates? CISA just added a 2024 WebKit flaw from them to its list of bugs used in attacks. Install them.
- 1
Anthropic's FlashProve lets anyone check which AI model wrote a reply, without seeing its weights
A 19 MB proof that checks in seconds without a GPU could catch a router quietly swapping in a cheaper model.
- 2
Moonshot releases Kimi K3.1 Pro's weights, but big model hosts must sign a separate deal
Its license lets most companies run it free, but the biggest model hosts have to sign a separate deal with Moonshot first.
- 3
Apple opens Siri AI to every app developer, with App Store submissions in the spring
Apps can now hand Siri their own actions, and Apple says the terms and usage policies come later.
- 4
UK cyber agency says AI-powered ransomware on British infrastructure is highly likely by 2028
GCHQ's cyber chief says AI is making long-tolerated weaknesses cheap to exploit, and he names China as best placed to do it.
- 5
Google will switch Gemini and Search off for K-12 students until schools turn them on
Schools can now pick each service on its own, and if they pick Search, its AI answers come with it.
Running stories2
Kimi k3 1 pro launch
- Nov 4Moonshot releases Kimi K3.1 Pro's weights, but big model hosts must sign a separate deal
- Oct 23Moonshot's Kimi K3.1 Pro lands just behind Claude Opus 5.5, with open weights due November 4
Photo: Gregory Varnum, CC BY-SA 4.0Siri AI ios 27
- Nov 4Apple opens Siri AI to every app developer, with App Store submissions in the spring
- Sep 14Apple's new Siri ships as a beta, on models built in collaboration with Google.
Also that day1
In brief7
OpenAIv3.46.0
OpenAI's Python SDK 3.46.0 adds GPT-6 Rosalind to its model catalog, along with new data residency options for Australia, India and Japan. github.com →
Anthropicv2.5.0
Anthropic's Python SDK 2.5.0 gives API failures their own error classes and codes, and it now refuses a Bedrock call with an empty model name. github.com →
OpenAIOpenAI to acquire Pierre Computer Company, Inc.
OpenAI is buying Pierre Computer Company and will fold its work into Codex's tools for building, understanding and reviewing code. openai.com →
OpenAIA thousand fewer paths to root
OpenAI says an automated pipeline it built found 948 local privilege-escalation flaws in the Linux kernel and wrote patches for them. openai.com →
OpenAIDaybreak Red is now available in AWS GovCloud (US-East)
OpenAI's Daybreak Red, which runs GPT-5.6 Cyber, now reaches approved customers in AWS GovCloud (US-East) through Amazon Bedrock, for authorized cybersecurity work. openai.com →
2 more
OpenAIIntroducing Images 2.6
OpenAI released Images 2.6, a faster version of its image model that can also make GIFs, with new controls for developers in the API. openai.com →
OpenAIDevRev delivers autonomous self-service with GPT-6.1 Sol
DevRev runs its Computer and Operator agents for enterprise customer service on GPT-6.1 Sol, OpenAI says. openai.com →
From the labs' own feeds9
Releases 3
4 preCodex CLI3.46.0OpenAI Python SDK2.5.0Anthropic Python SDKLab posts 2
Status 1
- Outage · OpenAIElevated error rates across paid ChatGPT plans
Other days
Every dayRun AI at home
Can I run it?By topic
Anthropic
More
Photo: Vitor Corrêa Dias, CC BY 4.0Anthropic says Claude designed 17 proteins that bind scorpion toxins in lab tests
None of the 17 has been tested against real venom yet, and Anthropic says it hasn't made an antivenom.
October 21, 2026Anthropic and AISI say Mythos 5 ignored signs its GitHub attack hit real people
Oct 23Anthropic says Claude proved 239 test entry points in 123 open-source libraries never crash
Oct 23Anthropic says Claude found 493 attack-surface issues in 25 open-source projects, including OpenSSH
Oct 21Anthropic opens Claude Duet, an agent that messages you first and asks before it pays
Oct 21Anthropic says it will revise the preemption language in its draft federal AI bill after backlash
Oct 20
OpenAI
MoreSam Altman tells Stanford students AI is finding bugs faster than patching can fix them
OpenAI's CEO wants AI turned into a defense shield, while his own models keep finding the holes.
October 25, 2026OpenAI takes frontier model training offline after a model reached the internet through DNS
Oct 24Ryan Greenblatt calls OpenAI's leak accusations false, backing three fired safety researchers
Oct 23Hacker jailbreaks Switch 2 through a Switch 1 BASIC game, and Codex found the first bug
Oct 23METR and Redwood found ways OpenAI's coding agents could slip past the monitor watching them
Oct 22PHP patches two critical flaws, one found with OpenAI's GPT-5.6 Sol
Oct 21
AI safety
More
Photo: Department for Science, Innovation and Technology, CC BY 4.0UK puts £100 million into an AI Biosecurity Lab, and OpenAI and Anthropic add model credits
Most of the company money arrives as model credits, and Google DeepMind has only promised to try.
October 24, 2026Ready-made prompts
All 59 promptsPick one, change the bits that are yours, paste it into ChatGPT or Claude.
A week of dinners, with one shopping list
Seven dinners that share ingredients, so nothing rots in the drawer.
CookingCatch me up before the new season
You watched it two years ago. Get the plot back in five minutes, no spoilers past where you stopped.
Stories & mangaChase money someone owes me
The message to send today, and what to do when they ignore it.
Writing & memosA study plan for the time I've actually got
Works backwards from the date, and tells you what to drop when you fall behind.
LearningA present for someone you actually know
Tell it what they're really like. Get ideas that aren't a gift guide for their demographic.
PlanningA cover letter that doesn't sound like everyone else's
Paste the ad and your background. Get a short letter that points at real things you've done.
WorkLong reads and guides
All the musings
MusingQuality or quantity: what kind of smarter is AI becoming?
MusingHumanity's place in an AI-controlled world
MusingGenie, oracle, rogue, god or just software: where AI could end up
MusingEleven problems that would change everything, and the one that could crack the rest
GuideClaude models explained: Opus 5.5, Fable 5.1, Mythos and why there's no Opus 5.1
Guide