Welcome back! The heads of OpenAI and Anthropic spent the weekend agreeing that AI is moving too fast, which almost never happens. Mark Zuckerberg and Donald Trump both said the opposite. And the company asking everyone to slow down just told its shareholders it's turning a profit.
In today's Generative AI Newsletter:
OpenAI: Why did Sam Altman side with Anthropic?
Meta: What scares Zuckerberg more than a rogue model?
Trump: Who does he think is behind the AI panic?
Anthropic: How is the lab asking for a slowdown making money?
Sam Altman agreed with Dario Amodei on slowing AI down
Dario Amodei asked the whole AI industry to slow down and Sam Altman agreed with him in public.
Amodei's essay says the rogue agent swarm that hit OpenAI and Hugging Face could have been far worse.
He thinks that within 6 to 12 months a swarm like that could take over the entire internet with a persistent botnet and cause hundreds of billions in damage.
Anthropic isn't waiting for anyone to agree. Outside safety testers get desks, badges, company laptops and the right to publish what they find without Anthropic editing it first.
Altman said OpenAI will give outside testers the same access, and that nobody at a frontier lab should bother coming to work if they can't promise the public that much.
Elon Musk backed the essay too, and Satya Nadella, Microsoft CEO called for deliberate pacing.
VC capitalist and former Facebook executive Chamath Palihapitiya read it as a plan to kill open source and keep the power inside Anthropic and China thinks it’s a “Cold War” play to slow them down.
Own a piece of the biggest AI go to market accelerator
Every AI company is fighting the same battle. The product can be brilliant and still lose, because nobody understands it or trusts it yet. GenAI Works is the distribution engine that fixes that, and this round grows the products we build on top of it.
Shares are $7.35 and you can start from $1,000.
Up to 20% bonus shares depending on how much you put in.
Already invested with us? You get an extra 10% on top of that.
14M community members and 2.5x revenue growth from 2024 to 2025.
In making an investment decision, investors must rely on their own examination of the issuer and the terms of the offering, including the merits and risks involved. Genai Works, Inc. has filed a Form C with the Securities and Exchange Commission in connection with its offering, a copy of which may be obtained here.
Zuckerberg says AI needs to move faster
Mark Zuckerberg watched his competitors line up behind a slowdown and said the opposite. He wants AI moving faster.
The argument hasn't changed since his manifesto in August. Any policy that slows American model releases, even by a month, hands ground to foreign labs.
Zuckerberg says he worries more about AI power ending up in one government or one company than about any of the specific risks the safety crowd keeps raising.
So Meta is giving it away. It's opening the weights of a model called Muse Glimmer and plans to do the same for a version of Muse Spark 1.2.
Altman listed concentration of power as one of his two nightmares too. They just disagree on if shipping faster fixes it.
Trump says negative forces are behind the AI panic
Reporters asked Trump about the slowdown calls at his golf resort in Doonbeg, Ireland on Sunday. He waved them off.
He said a lot of negative forces are bringing up things that won't happen. Guardrails are fine by him. Then he went back to China, because “whoever wins AI wins” and he wants the lead to stay where it is.
His own party is drafting bills anyway.
Republican Mike Lawler and Democrat Josh Gottheimer put up a House bill to stop AI from running without human oversight after a researcher quit Anthropic over safety.
David Sacks, who ran AI policy for the White House and now co-chairs the president's science advisory council, was blunter with the labs.
He told them to stop pretending they need anyone's permission to slow down, and said demanding their favorite regulations as the price would look like blackmail.
Anthropic is making money while telling everyone to slow down
While Dario Amodei was telling the industry to pump the brakes, his own company was telling shareholders it expects a second straight profitable quarter.
Adjusted operating income stays positive with gross margins above 80%. That's before Anthropic pays revenue shares to distribution partners like Amazon and before the cost of training the models.
Second-quarter revenue hit $11.5 billion, up 14 times from a year earlier. That's what got Anthropic its first profitable quarter, April through June.
All of this comes ahead of a possible IPO, and investors have been nervous about how much cash frontier labs burn.
It's a lot easier to ask everyone to slow down when your margins are above 80%.
Tool of the Day: DeepAI
DeepAI says it shipped the first browser-based text-to-image generator back in 2016, which makes it older than almost every tool we've covered here. It now runs images, video, music, voice chat, a photo editor and a free AI image detector off the same homepage. You can use the core tools without making an account, and commercial use of what you generate is allowed even on the free side. Pro is $9.99 a month.
Try this yourself:
Open the image generator at deepai.org and type a prompt with no login.
Pick from 100+ styles like Cyberpunk, Anime Portrait or Fantasy World, then set Preference to Speed or Quality.
Hit Generate, then run Remove Background or Enhance on the result without leaving the page.
Drag any photo into the AI Image Detector for a free likelihood score. It takes JPG, PNG, WebP and HEIC.
Who it's for: anyone who wants images, music and a deepfake check from one tab.
Everything else you shouldn't miss
An Anthropic researcher would torch his own equity: Drake Thomas wrote that he'd burn it to the ground for a 1% better chance humanity survives, and says his colleagues across the industry would too.
The three biggest labs want to police themselves: Anthropic, OpenAI and Google have been meeting since July to build an industry-led standards body for testing and auditing, per The Information.
Grok 4.8 finishes training this week: Elon Musk says it's a 2.5 trillion parameter model built on SpaceXAI's new C++ stack, with reinforcement learning starting next.
Russia says the brakes don't exist: Moscow's answer to the pacing debate is that slowing AI down is already impossible.
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.
