👋 Welcome back to AI for SME Success, your weekly dose of practical AI insights that matter to small businesses.
Today we cover:
- Why small businesses choose Anthropic
- Why you should monitor AI Overviews of your business
- 8 AI use cases where hallucinations won’t cost you much
- NotebookLM updates, plus Visa’s move into ChatGPT payments
- When AI argues against itself in court, a vendor management lesson
Let’s dive in! 👇
💪 The Most Capable AI Model, Suspended in Days
Why small businesses choose Anthropic.
On June 9, Anthropic launched Fable 5, the most capable model it has ever released to the public, outperforming rival models across leading benchmarks.
On June 12, citing national security, the US government barred foreign nationals from both models. To comply, Anthropic disabled Fable 5 and Mythos 5 for all customers.
To put this into perspective, Anthropic was founded in 2021 by seven former OpenAI leaders who left the company due to concerns over its increasingly commercial direction and insufficient safety controls.
The bet on safety and quality over speed is now winning customers. According to Apptopia, Anthropic’s market share jumped from 2% in Sep 2025 to 17% in Oct 2026.
US gen AI chatbot market share by daily active users. Source: Apptopia.
Similar to many other users, I ended my ChatGPT subscription earlier this year and started using Anthropic. Reasons I chose Anthropic:
- Clear leader in developing the most capable AI.
- Built on a foundation of safe and responsible AI.
- Business-oriented right from the beginning.
- Offers a dedicated small-business path.
🤥 AI Overviews Can Lie About Your Business
Regularly search your own business name in AI.
Google’s AI Overview falsely tied two Munich publishers to scams and shady dealings. None of the claims appeared in the linked sources.
The AI summary is what most people see. Pew Research found only 1 percent of users click a source link inside an AI Overviews.
And these summaries are often wrong. Earlier this year Oumi’s audit for the New York Times found that while Gemini 3 overviews were accurate 91 percent of the time, 56 percent of those correct answers cited sources that did not actually support them.
On May 28, a German court became the first to hold Google directly liable for false claims its AI Overviews invented about a real business. The court ruled that AI Overviews are Google’s own statements, not a list of links.
Takeaway: Most small businesses focus on getting their name into AI tools and overviews. The bigger concern is getting it right. Regularly verify whether AI represents your business accurately.
🌱 AI Use Cases in Farming
8 AI use cases where hallucinations are low risk
OpenAI’s ChatGPT Pro Community recently profiled Hiroki Tomiyasu, a broccoli farmer in Hokkaido who taught himself to farm and then taught himself to automate it. Here are eight ways he puts AI to work on his farm:
- Diagnoses crop disease from a photo
- Integrates satellite data into crop monitoring
- Adds technical notes to images
- Builds remote control for greenhouses
- Creates a custom group chat for workers
- Counts plantings from group chat history
- Researches buy-versus-build decisions
- Builds a farm management database
Takeaway: each of these is a spot where a wrong AI answer is cheap to catch or cheap to fix, so hallucination risk stays low. Good places to start.
Note: on June 12, 2026, OpenAI launched three new OpenAI Academy courses to help people apply AI at work. They build in sequence: AI Foundations, Applied AI Foundations, and Agents and Workflows.
🛠️ NotebookLM Updates and Visa AI
NotebookLM Updates (Google)
NotebookLM can now write and run code in a secure cloud computer, using 100+ software skills for deeper research and complex analysis. It generates downloadable, editable outputs in many formats (PDF reports with charts, spreadsheets, slides, images, structured data), and helps build your source library from loose ideas using Google Search.
Visa and OpenAI
On June 10, Visa embedded its payment network into ChatGPT, so an agent can buy at almost any merchant that accepts Visa. The system replaces Instant Checkout and includes built-in guardrails such as tokenization, strict spending caps, approved merchant restrictions, and manual confirmation.
🫢 When AI Argues Against Itself in Court
Your suppliers use AI too. Control what they hand you.
We’ll close this edition with a story that illustrates where AI use sits right now.
A US federal judge canceled a trial and sanctioned four lawyers after attorneys on both sides of a fee dispute in Aberdeen, Mississippi filed briefs citing cases that did not exist. Two were barred from the district court for two years. One argued she did not know AI could invent cases. The judge called that “insufficient and incredulous” and sanctioned her anyway.
Here’s what this story tells us: AI is now in the hands of highly respected professionals, and many of them do not know how to use it safely. One of those people could be your supplier, your lawyer, or your accountant.
Takeaway: Add restrictions on AI use to your contracts and require vendors to disclose any AI-assisted work.
Thank you for reading today’s edition!
If this issue was valuable, pass it along to a fellow business owner. I’d love to hear your feedback at natalia@nataliabrattan.com.
See you next week,