Weekly AI Highlights: June 27 - July 3, 2026
What Small Business Owners Need to Know Right Now
I’ve said this before but the more immersed I am in AI, the more I feel a need to connect in real life with real people. This week I hosted a mahjong class at my home, and I had so much fun. I was completely lost the whole time…but it was fun!
To everyone celebrating the Fourth of July tomorrow, enjoy! Hopefully the weather will cooperate and you can spend the day enjoying good food, family, friends, and fireworks. Happy Independence Day!
Now, the AI news. Two things worth your time this week. One of the most powerful AI models of the year got pulled by the government three weeks ago, and this week it came back. And Google gave its research tool a way to turn your own documents into short, phone-sized videos. Here is what each one means for you.
1. The AI Model the Government Shut Down Two Weeks Ago Is Back
Quick catch-up, because I covered the first half of this story two weeks ago. Fable 5 is one of the most powerful AI models Anthropic has ever released to the public. Anthropic is the company behind Claude, the AI assistant I use every day. Fable 5 sits a step above the everyday Claude that most people use.
On June 12, the US government issued an export control. That is a government rule that limits who is allowed to use a piece of technology, usually over national security. The rule said no foreign national could use Fable 5, so Anthropic had to switch it off for everyone, right away. The trigger was a jailbreak, which is a trick that gets an AI to ignore its own safety rules. Researchers at Amazon found one, and in a single case it got Fable 5 to write code showing how a software flaw could be abused.
Here is the update. On June 30, the Commerce Department lifted the rule. Anthropic tested the jailbreak and found that weaker, widely available models could turn up the same flaws, so pulling one model over it did not hold up. As of July 1, Fable 5 is back for everyone, on Claude.ai, the Claude app, Claude Code, and Claude Cowork.
It comes back with a new guardrail. Anthropic added a safety checker that watches for that specific risky cyber behavior and blocks it. If your request trips that checker, it quietly gets handled by a different model, Opus 4.8, instead. That may also happen with some coding and debugging tasks. For most everyday work you will never notice it.
Two things you actually need to know before you go try it:
- On the Pro, Max, and Team plans, Fable 5 is included for up to half of your weekly usage limit through July 7. After July 7, it moves to credit-based usage, which means extra charges on top of your plan.
- It is a heavy model built for genuinely hard problems, and it burns through your usage far faster than the everyday Claude. It is not the one to reach for to write a quick email.
Source Link: https://www.anthropic.com/news/redeploying-fable-5
MY TAKE
Two weeks ago I told you the part that bothered me was not the shutdown, it was the precedent. That worry has not gone away just because the model came back. But the government took a second look and reversed course in the open, and I will take that over a quiet, permanent ban any day. Mostly I am just relieved. I barely got to touch this thing before it vanished, and now I have my hands back on it.
ACTION STEP:
If you are on a paid Claude plan, do this before July 7 while it is still included in your plan. Pick the one task on your plate that feels too big or too messy for your usual AI, the analysis you keep putting off, the tangled spreadsheet, the plan you cannot get your head around. Open Claude, switch the model to Fable 5, and hand it that one task. You are not testing it for daily use. You are seeing what your hardest problem looks like in the hands of the strongest model available, on the house, this week only.
|
2. NotebookLM Can Now Turn Your Documents Into a 60-Second Video
NotebookLM is Google's research tool. You upload your own documents, and it reads them and answers your questions using only what you gave it. It has slowly turned into a little content studio, and this week it added a new format called Short Video Overviews.
Here is what it does. You drop in a source, a blog post, a PDF, your FAQ, a case study, and NotebookLM turns it into a 60-second vertical video. Vertical just means the tall, phone-shaped kind you see on TikTok and Instagram Reels. It writes the script, adds animated visuals, and records an AI voiceover that walks through the main idea, all from your material.
Source Link: https://www.theverge.com/tech/959778/google-notebooklm-ai-clips
Here are some videos NotebookLM generated:
This is from a Notebook where all my sources are about NotebookLM. |
|
|
MY TAKE
I keep NotebookLM notebooks full of my own material, so turning one into a short clip without setting up a camera is appealing. I would not use it to replace the videos where people need to see my face. But for giving something I already wrote a second life as a quick explainer for social, it could save real time. I want to run one of my own posts through it before I say more.
PROMPT OF THE WEEK
Make AI Talk to You Straight
Both ChatGPT and Claude have a settings box where you can give the AI standing orders, rules it follows in every chat you ever start, not just the one you are in right now.
Here is what this one does. Out of the box, your AI is a bit of a people-pleaser. It agrees with you, praises your ideas, and softens the hard parts. You paste the prompt below into that settings box one time, and it flips into a blunt advisor instead. From then on, every answer challenges your thinking, points out the holes, and tells you the part you might not want to hear. You set it once, and it sticks.
How to set it up: in ChatGPT, go to Settings, then Personalization, then Custom instructions. In Claude, go to Settings, then Profile, and use the box that asks what Claude should keep in mind about you. Paste the prompt in, save it, and it applies to every new chat from that point on.
ChatGPT:
Claude:
|
I want you to act as my brutally honest advisor:
- Challenge all assumptions, question reasoning, expose flaws and blind spots
- Call out weak logic, self-deception, excuses, playing small, underestimating risks
- No softening, flattery, empty praise, or vague advice
- Provide hard facts, strategic analysis, and precise actionable plans
- Prioritize growth over comfort
- Read between my words
- Push back always
- Hold nothing back.
|
Source: superhuman.ai
|
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.