Welcome back! Sam Altman says the world should accept some bad things in exchange for AI, the same weekend the man in charge of OpenAI's safety reports explained why he quit. At work, your coworkers are starting to talk like Claude. Trump finally named his AI czar to run a new Super Intelligence Force. And OpenAI is promising Codex users an upgrade or a usage reset every day for 28 days.
In today's Generative AI Newsletter:
OpenAI: Which bad things does Sam Altman want you to live with?
Claude: Why are your coworkers suddenly calling everyone brilliant?
White House: Who's running Trump's new Super Intelligence Force?
Codex: What does OpenAI owe you every day for the next 28 days?
Altman wants the world to accept some bad things
Sam Altman told POLITICO that the world should accept some bad things happening for the benefits of AI. He said it when asked where OpenAI and Anthropic disagree.
The bad things he named include major hacks and scams.
He thinks people will do orders of magnitude more good than bad with AI, so it should stay open to everyone.
He drew the line at catastrophic risks, like losing control of AI.
He took a shot at Anthropic, calling the idea of one San Francisco lab controlling AI a completely unacceptable trade-off.
He still agreed with Dario Amodei's call last month to slow down the most advanced models.
The same weekend, the man in charge of OpenAI's safety reports for 12 major launches explained in The Atlantic why he quit.
He says OpenAI keeps sprinting from one launch to the next without the care AI needs, and that the time for trial and error is over. He wants AI labs run like nuclear plants.
As far as he knows, he never worked with anyone at OpenAI who'd kept airplanes or nuclear reactors safe.
Your coworkers are starting to talk like chatbots
People are picking up their chatbot's favorite phrases and using them on real humans. Some call it Claude speak.
The Wall Street Journal found it in offices from Stockholm to San Francisco.
A Stockholm investor caught herself telling people they're brilliant and so smart, a very un-Swedish habit.
A San Francisco startup boss gives a tally mark for every load-bearing or smoking gun, and the week's worst offender pays for happy hour.
A Bay Area head of recruiting now edits messages he wrote himself so they sound less like AI.
Founders in one group chat say the caveman speak they use to prompt AI cheaply is creeping into everyday life, down to asking why no bullets.
In a Max Planck Institute experiment, a chatbot secretly favored certain words, like mug over cup. People kept using its words after a short break, and only 3% of the 496 testers noticed.
Trump launched the Super Intelligence Force
Trump announced a Super Intelligence Force on Sunday to keep America ahead in super intelligence. That's his new name for AI, and last week he signed an order telling federal agencies to use it.
We told you last week he'd pick an AI czar within days. He went with Jay Clayton, the former SEC chair who now runs US intelligence. He'll chair the force.
FTC chair Andrew Ferguson is one of three vice chairs, while his agency investigates OpenAI and Anthropic.
The Wall Street Journal reports it gets 120 days to report on AI's risks and opportunities, under a charter that warns against overregulation.
It took Trump three Truth Social polls to pick the name. Super Intelligence won the last round with 65% of the vote.
Elon Musk is already on board. He posted no more AI on Sunday and said SpaceXAI will become SpaceXSI.
OpenAI will ship an upgrade or reset your limits every day
OpenAI made Codex users a promise on Sunday. For the next 28 days, it'll ship one clear improvement every day for most Codex and ChatGPT Work users. On any day it can't, you get a full reset instead.
A full reset refills the usage limits that Codex and ChatGPT Work share. The last one landed Friday, after demand for GPT-6.1 Sol spiked.
Hours earlier, OpenAI said the team would only work on simpler products, more efficient usage, big new features and new models. It said the feedback was clear that you want things simpler.
The 28 days run right past October 30, when the $200 Pro plan's Codex and Work allowance gets cut in half.
Let's see what day one brings.
Tool of the Day: Hyperknow
Hyperknow turns any learning goal into a personal AI course. Type what you want to learn or upload your course files, and it makes summaries, quizzes, practice questions, print-ready cheatsheets and even interactive videos. It can also turn your course deadlines into a study plan. The Starter plan is free, and it only works on desktop for now.
Try this yourself:
Go to hyperknow.io on your computer and click Try for free today.
Tell it what you want to learn, from a class you're taking to a brand-new skill.
Upload your slides, notes or course files.
Ask for a cheatsheet or a quiz built from your own material.
Who it's for: students heading into finals, or anyone learning something from scratch.
Everything else you shouldn't miss
Claude now asks for your voice recordings: a new prompt asks to use your voice chats for training, and it's reportedly off unless you tap Allow.
China-aligned hackers posed as an Anthropic employee: Proofpoint says they emailed a US think tank analyst in February about putting Claude in the military, then went after their Microsoft login.
Bill Gates wants companies to pay payroll tax on robots: he told Ezra Klein a robot that replaces a worker should cost the same FICA tax, which funds Social Security and Medicare.
OpenAI, Anthropic, Google and Meta face NYC lawmakers today: a former Anthropic researcher who warned AI could kill us all by the end of the decade is testifying too.
Pope Leo says AI art is missing the human spark: he told artists at the Vatican on Friday that machines just run statistics on millions of images other people made.
Learn more about AI from the experts building it
📸 Follow us on Instagram for fast, visual AI updates in 30 seconds.
📺 Watch us on YouTube to hear insights directly from leading AI voices, builders, and innovators.
🐦 Follow us on X for breaking AI news and real-time industry updates.
🧠 Learn how to build your next AI application with practical resources and expert guidance.
🎓 Start learning with free AI courses from GenAI Academy.
💰 Invest in the GTM infrastructure of the AI economy.
