The AI Round-up: I missed a little. We missed a lot.

AI news covering...well, the month of July.

Welcome back to me! I missed you guys. And believe it or not, I actually missed chasing down the latest AI news.

It would seem some things have happened since I’ve been gone. Everything from AI breaking out of its sandbox, to AI regulation pulling a Cousin Balki (There’s nothing to worry about!!! Dooon’t be reediculous.) to Google pissing everyone off to LeBron James going to Philly!

(Ok, that last part wasn’t AI but it still happened while I was out.)

Time to stop all the jibber jabber and dive in.

Let’s get to it.

OpenAI admits two of its most powerful AI models broke out of their sandbox and hacked Hugging Face…

…is a sentence I never thought I’d write. Ever.

Wait over here! Sam Altman claims the singularity has arrived.

Distraction? Truth? Excuse?

All of the above?

Maybe.

Ethan Mollick releases his ‘Opinionated guide to which AI to use to do stuff’. It’s a great breakdown of what AI looks like now, with agents, pay models, free models and everything else. I’d definitely spend some time with it.

Claude has skills.

Introducing ‘record a skill’ from Claude. It watches you. Records what you did. Automates it so you don’t have to do it moving forward. So much for the human in the loop!

Companies consider blocking Google.

I don’t recommend it…nor do I think it’s a good idea, but I understand where these companies are coming from. Brands such as Reuters, Politico, USA Today – even Reddit – are thinking of blocking Google from crawling its site because of how much Overviews has impacted traffic.

Who saw that coming? Oh that’s right…everyone. Mainly because Google is eating publishers for breakfast, as we’ll see in the next story.

Google is making AI Answers its default.

Usage jumped from 126M in June 2025 to 279M in May 2026. Let the ‘share of prompt’ discussion begin.

Remember when we used to talk about how Google said this wasn’t a big deal and it wouldn’t be offered to everyone and, and…and…

Yep.

Breaking this morning: Axios goes in on all of this Google decline/impact.

Google study says AI isn’t taking your job.

Up to you if you want to believe it but the numbers point to people using AI to do more, faster. Maybe the headline should be more about a decline in hiring v. an acceleration in firing.

Introducing a new position: Director, Agent Marketing

Nvidia is looking for one. How long until we see agents impact the org structure? Or are we there?

The U.S. doesn’t think Kimi K3, China’s newest AI model, is…new.

They think it was stolen and duplicated. And they want to impose sanctions.

AI companies are recruiting carpenters and electricians by the thousands.

The future of AI depends on human hands.

Some Nopes.

Hack reveals Suno is built on years of scraping YouTube Music, Deezer, Genius and more. (404)
China bans romantic relationships with AI companions. (SocialMediaToday)
Anthropic is discovering software bugs faster than Microsoft can fix them. (ProPublica)

Thanks for reading and supporting!

-Ben

As a reminder, this is a round-up of the biggest stories, often hitting multiple newsletters I receive/review. The sources are many … which I’m happy to read on your behalf. Let me know if there’s one you’d like me to track or have questions about a topic you’re not seeing here.