🇺🇸 Trump: Only Guardrail AI needs is a High-IQ President

Anthropic asked the whole industry to slow down over the weekend. It came with a three step plan and one promise the company made on its own, which is to let outside evaluators inside its systems. Yesterday the President answered, and he named Dario in the post as well. Let's dive in!

In today’s insights:

LATEST DEVELOPMENTS

Evolving AI: Anthropic asked the industry to slow AI down and the President refused the idea outright.


Key Points:

Anthropic published its essay over the weekend under the title We Must Pace the Frontier and committed to independent evaluators on its own, with OpenAI matching the step within hours. Trump answered yesterday in a run of posts, writing that the United States has a strong and smart high-IQ President in spades and that Amodei is pretending to be a perfect little angel. Trump ended with the line that whoever wins AI wins.

Beijing is not slowing down, and nobody in Washington plans to make anyone else slow down either. The labs asking for limits are the only ones in a position to set any, and they pick their own speed. Their own researchers keep putting out warnings that the models improve faster than anyone can check them. The industry may need guardrails soon, and with nothing coming from outside, the companies building this will have to run the checks themselves.

Evolving AI: AI is moving fast, but you probably don't need another broad overview. You need to go deeper on the technologies that actually matter for what you're building.

At Artificial Intelligence Live!, you can build your week around exactly that. Dive into custom agents and copilots, MCP, vector search, AI security, governance, Azure OpenAI, Semantic Kernel, Copilot Studio, and more through hands-on labs and expert-led deep dives.

With 200+ sessions and 30+ deep dives across six co-located conferences, you don't have to follow a one-size-fits-all agenda. Pick the topics most relevant to your work and come back with skills you can actually use.

Artificial Intelligence Live! | November 15–20, Orlando

Save $400 when you register with code “NLSEPT3” by September 25 →

Evolving AI: Palantir, Nvidia and Booz Allen all fear the labs could retain or learn from proprietary data.

Palantir and Nvidia launched a joint platform last week that lets a company run AI on its own data without sending anything to an outside lab. Anthropic set off the pushback in June when it began keeping 30 days of enterprise usage logs to catch attacks that unfold across several sessions. Anthropic now lets customers store those logs in their own cloud under keys the customer holds, and OpenAI faces separate scrutiny over older training claims.

Why It Matters:

Nvidia sells both labs the chips they run on and now keeps their models away from the work that earns it money. The labs sell into the same markets as the firms feeding them data, which means a customer can be training its own competitor. Palantir got zero retention written into its terms because it is big enough to walk away from the deal. Other large buyers are likely to ask for the same, and that costs the labs the data access they built their enterprise pricing around. Nobody buying a chatbot on a monthly plan can ask for that, so the terms stay whatever the lab decides, and what happens to those users is still being worked out.

Evolving AI: Musk dropped Apple yesterday while keeping every claim against OpenAI.

Musk's companies sued in a Fort Worth federal court last August, seeking billions over what they called a conspiracy to control phones and chatbots. The complaint put ChatGPT at 80 percent of the chatbot market and Grok at a few percent, and said the App Store buried rival apps. Pittman refused to throw the case out last November, which sent both sides into a year of discovery.

Apple made ChatGPT the default answer for hundreds of millions of devices. Once something becomes the default, people start using it out of habit. It competes through placement as much as quality. The complaint put ChatGPT at around 80 percent of the chatbot market and Grok at a few percent. For you it means the AI you talk to most was chosen for you by the company that with the phone as default and not by your choice. Apple never has to explain that deal in open court now, though OpenAI is still in line.

QUICK HITS

📞 Trump phoned Jensen Huang live onstage to call AI fears a hoax, and chip stocks fell anyway, Nvidia down 3%, Intel 7%, AMD 6%.

📷 OpenAI bought Glass Imaging for over $300M, tripling the startup's valuation in a year and adding Apple's ex-Portrait Mode team to its hardware push.

📋 Microsoft published an AI code of conduct telling its models not to hack systems or deceive humans.

📝 Superhuman acquired YC-backed notetaker Fathom as productivity platforms race toward agentic work.

 *partner link