Weekly AI Highlights: June 13 - 19, 2026
What Small Business Owners Need to Know Right Now
I spent this past week in New York onboarding a new client, catching up with a few friends, hanging out with my family, and witnessing a whole city losing its mind over the Knicks. They won their first title in 53 years, and yesterday's parade turned Manhattan into one giant block party. What a week to be here.
Here is the AI story I could not stop thinking about. Fable 5 came out last week, the most powerful model Anthropic has ever released to the public. I had big plans to spend the weekend putting it through its paces.
Then days later, the government made them pull it. So much for that. I barely got to touch it.
1. The Government Ordered Anthropic to Shut Down Two of Its Most Powerful AI Models
Anthropic is the company behind Claude, the AI assistant. Fable 5 and Mythos 5 are its two most powerful models, a step above the versions most people use day to day.
On June 12, the US government issued what is called “an export control directive.” That is a government order, usually tied to national security, that restricts who is allowed to access a piece of technology. This one said no foreign national could use Fable 5 or Mythos 5, whether they were inside or outside the United States, including Anthropic's own foreign-national employees. To comply, Anthropic had to switch both models off for every customer, right away.
The government's concern, as Anthropic understands it, is a "jailbreak." A jailbreak is a trick that gets an AI to ignore its built-in safety rules. Anthropic says it reviewed the specific technique and found it only surfaced a few minor, already-known flaws, the kind other widely available AI models can find too, and that no actual harm has been demonstrated.
Anthropic is complying with the order but strongly disagrees with it. Its argument: pulling a model used by hundreds of millions of people over one narrow, unproven jailbreak would, if applied across the whole industry, effectively halt every new AI release. The company says the government should be able to block genuinely unsafe AI, but through a process that is transparent, fair, and grounded in technical facts, and it believes this is a misunderstanding it is working to reverse.
One thing worth knowing: every other Claude model is unaffected. If you use Claude day to day, your tools still work.
Source Link: https://www.anthropic.com/news/fable-mythos-access
MY TAKE
What gets me is not the shutdown itself, it is the precedent. If a single unproven jailbreak can pull a model used by millions, that is a standard no AI company could survive. I do think the government should be able to step in when something is truly dangerous. I just want that line drawn out in the open, with real evidence, not a verbal heads-up at the end of the day on a Friday.
2. ChatGPT Now Has a Home for Tasks You Schedule in Advance
Here is something you may not know ChatGPT could do: you can ask it to do a task later, or on a repeating schedule, instead of right now. Think of it as a standing instruction that actually does the work, not just a reminder that nags you.
This week OpenAI made that feature much easier to find and use. Open the ChatGPT sidebar and you will see a new page called Scheduled. It shows every task you have set up, when each one is due to run, and lets you pause, edit, or delete any of them in one place. OpenAI also says the tasks now run faster and more reliably. You can set a task for an exact time or a looser window like "every weekday morning."
There is also a monitoring option, where ChatGPT checks the web or your connected apps on your behalf and reports back.
One note on access: it is rolling out to the paid plans (Go, Plus, Pro, Business, and Enterprise) on web and mobile. No word yet on the free version. OpenAI is also retiring Pulse, its old daily-summary feature, and folding that job into this new scheduling page.
Source Link: https://help.openai.com/en/articles/6825453-chatgpt-release-notes
MY TAKE
I've been using scheduled tasks for a while, so for me this is not a new feature, just an easier one. What I love is that I can finally SEE everything in one place.
ACTION STEP:
If you are on a paid ChatGPT plan, set up a daily brief this week so the news comes to you instead of you chasing it. Open ChatGPT, paste the prompt below, and it will create a scheduled task that runs every morning on its own. Just swap in the topics you care about:
“Every morning at 7am, give me a short daily brief on the topics I care about most: [your topics here, for example AI tools for small business owners, news in my industry, and what my main competitors are doing]. Keep it to the 5 most important updates, one sentence each.” After that, it lives on the new Scheduled page, where you can pause or change it anytime.
|
3. Claude's Design Tool Just Moved Closer to Competing With Canva and Figma
Claude Design is Anthropic's tool for making visual things by describing them. You type what you want: a slide deck, a landing page, a one-pager, a social graphic, and Claude builds it for you, then lets you adjust it by chatting or by editing it directly.
This week it got three upgrades. First, a more hands-on editor, so you can fine-tune a design yourself instead of only asking Claude to redo it. Second, a much longer list of places you can send your finished work, now including Canva, Adobe, Gamma, Miro, and build tools like Replit, Vercel, Lovable, and Wix, on top of the existing PDF, PowerPoint, and web-page exports. Third, tighter links with Claude Code, Anthropic's tool for turning a design into a real, working website or app.
A couple of honest caveats. Claude Design comes with the paid Claude plans (Pro, Max, Team, Enterprise), not the free one, and it is still labeled “in beta”, so expect some rough edges.
Source Link: https://claude.com/blog/claude-design-stays-on-brand-for-daily-work
MY TAKE
I already lean on Claude for so much that making a landing page or a set of social graphics without leaving it sounds appealing. I am not ready to close Canva, the templates and brand kit still win for quick polished assets, but for a fast first draft I can just describe, this keeps getting harder to ignore.
PROMPT OF THE WEEK
The Second Opinion
What it does: Turns any AI tool into a tough, honest editor for something you wrote with AI help. You paste in your draft and get back a checklist of what to fix before you publish.
When to use it: Right after you finish a draft (a post, an email, an article, a newsletter section) and before it goes out.
What you get: Four things in a checklist: what you might be missing, fresh angles you have not thought of, any claims that need a source, and one strong argument against your own point so you can see how solid it really is.
Works the same in ChatGPT, Claude, Perplexity, or Gemini. Paste the prompt, then paste your draft underneath.
|
Act as a sharp, honest editor giving me a second opinion. Below is a draft I wrote with AI help. Review it and give me a checklist with four parts:
1. What I might be missing: blind spots, places where I lean too far one way, and any viewpoint I left out.
2. Three angles I have not considered.
3. Any claims that need stronger evidence or a source before I publish.
4. One strong counterpoint to my main argument, so I can test how well it holds up.
Keep each item short so I can quickly decide what to keep, change, or research further.
Here is my draft:
[paste your draft here]
|
|
Comments
If any of this made you think differently about how you want to work with AI, hit reply and let me know. I read every message. Your questions shape what I cover next.
Have a great weekend,
Founder, Front Row AI Club
PS. Was this newsletter forwarded to you? Sign up to get it in your inbox.
You can find my previous newsletter articles here: The Front Row AI Report
© 2026, Front Row AI Club. All rights reserved.