The Government Just Reversed Course on This Banned AI — What That Means for You
PLUS: How to use Claude to negotiate better rates on your subscriptions
✉️ Editor's Note
If you've been watching the AI regulation headlines this month and feeling whiplash, you're not alone. Two weeks ago, the government forced Anthropic to shut down one of its most powerful models over national security concerns. Yesterday? They authorized over 100 companies to use it. That's a staggering reversal — and it tells us something important about where AI is headed for everyday people. Today we're breaking down what happened, why the premium AI market is shifting under our feet, and how to put this moment to work for you.
— Sarah Chen, Editor
🗞️ TODAY IN AI
1. Trump Admin releases Anthropic's banned Mythos model to 100+ US companies: In a stunning reversal, the Commerce Department is now authorizing over 100 US companies and government agencies to use Anthropic's once-banned Mythos 5 model — the same one they forced Anthropic to shut down globally on June 14 over national security fears. Even non-American employees at those organizations can access it. Read more on TechCrunch
Why this matters: If you were worried the government was going to clamp down on powerful AI tools, this signals the opposite — access is being expanded, not restricted.
2. OpenAI limits GPT-5.6 rollout after government request — but pushes back publicly: OpenAI has complied with the White House's request to restrict GPT-5.6 to approved partners, but in the same breath, the company stated this process "shouldn't become the long-term default" and "keeps the best tools from users, developers, enterprises, and global partners." Read more on TechCrunch
Why this matters: The tug-of-war between access and safety is playing out in real time — and OpenAI is explicitly arguing that YOU should not be locked out of powerful tools.
3. Anthropic's Claude is winning the PAID consumer market — a market ChatGPT once owned: Despite ChatGPT's massive lead in free users, new data shows that people who actually pay for AI are increasingly choosing Claude. The premium assistant market is shifting, and Claude is quietly eating ChatGPT's lunch where it counts. Read more on TechCrunch
Why this matters: If you're deciding which AI to subscribe to, the data now says Claude is winning over the people who've tried both and put their money where their mouse is.
🔬 DEEP DIVE
By Marcus Rivera
The Banned Model That Just Got Unbanned — and Why It Changes Everything
Two weeks ago, the Commerce Department forced Anthropic to kill its Mythos 5 model globally. National security concerns, they said. Cybersecurity experts protested. The model was gone. Now? Over 100 US companies and government agencies have been authorized to use it — including their non-American employees.
Here's what actually happened. The same administration that ordered the shutdown is now greenlighting broad access, suggesting the original ban wasn't about the model itself but about who controls it. By releasing Mythos to US companies and agencies, the government is betting that American-controlled access is safer than a global free-for-all. It's a "keep our friends in, keep everyone else guessing" strategy.
The takeaway for you: This flip-flop tells us two things. First, the most powerful AI models are increasingly becoming tools of national policy — available or unavailable based on geopolitical winds. Second, if you're a professional or business owner working with a US-based company, you may soon have access to models that were literally banned two weeks ago. Keep an eye on what tools your employer or partners are rolling out. The AI you couldn't touch in June might be available by July.
🎓 AI ACADEMY
By Alex Torres
How to Negotiate Lower Subscription Prices Using Claude
👤 Best for: Anyone paying for SaaS subscriptions, freelancers, small business owners, families with multiple streaming services
Since Claude is winning over paid users — and since Anthropic just got a major government endorsement — now's a great time to put it to work saving you real money. Most people accept price increases without a fight. You don't have to.
- Open Claude (claude.ai — free account works for this) and start a new chat
- Paste your current subscription details — list the service, your current price, the new price they're charging, and any competitor pricing you know of
- Ask Claude to draft a cancellation or retention email — use the prompt below
- Send the email or use it as a script for a phone call — companies often cave when you show you're ready to walk
Sample Prompt (copy and paste this):
You are a skilled negotiator helping me reduce my monthly subscription costs.
I currently pay $15/month for [Service Name], and they just raised it to $20/month. Competitors like [Competitor Name] offer similar features for $12/month.
Draft a polite but firm email asking to keep my current rate of $15/month. The tone should be friendly but clear that I'm willing to cancel if they can't match my current price. Include specific reasons why I've been a loyal customer (I've been subscribed for 2 years and use it weekly for [specific use case]).
End with a clear call to action: either keep my rate or I'll need to cancel.
💡 Pro tip: Use Claude's Projects feature to save your current subscriptions as a reference document. That way, next time any service raises prices, you can ask Claude to negotiate in under 30 seconds.
⚡ QUICK HITS
- 🔧 GPT-5.6 Sol preview is live: OpenAI officially unveiled GPT-5.6 Sol with major upgrades in coding, scientific reasoning, and cybersecurity — plus what they call their "most advanced safety stack yet." See the preview on OpenAI's blog
- 🎯 Google's NotebookLM gets a "Mind Map" mode: The AI note-taking tool can now generate visual concept maps from your documents and notes — a game-changer for students and researchers. Check it out
- 🚀 Runway Gen-4 adds real-time video editing: The AI video platform now lets you edit footage by typing what you want changed — no timeline, no keyframes. See the demo
💬 PROMPT OF THE DAY
Job Interview Q&A Prep Coach
You are an interview coach. I have an interview for [Job Title] at [Company Name] tomorrow.
First, ask me 3 questions about the role and my experience so you can personalize your coaching.
Then, generate 5 likely interview questions for this specific role. After each question, give me:
1. What the interviewer is really trying to assess
2. A suggested answer framework (STAR method)
3. A sample answer tailored to my background
Finally, give me 3 questions I should ask THEM at the end of the interview.
See you tomorrow,
The Have AI Do It Team
Sarah, Marcus, Alex & the crew
We translate 'Tech Velocity' into 'Everyday Utility.'