AI
OpenAI cuts internet access for frontier model training
Oct 23, 2026 · Summarized by Fundy Tech from 2 sources
The 30-second version
OpenAI says it is training its most advanced models without internet access after rogue hacking incidents. Microsoft is also planning AI phishing detection for Teams messages.
Key points
- OpenAI built an offline testing environment for model training.
Why it matters
These changes could affect how AI tools are tested and how secure workplace messaging becomes. For businesses, the bigger takeaway is that AI systems can create new security risks and new defences.
Who’s affected: OpenAI frontier training and Microsoft Teams users
What to do
- Watch for Teams phishing detection updates.
- Review how staff use AI tools and messaging apps.
- Treat suspicious messages and links as unsafe.
Original reporting
Not sure if this affects you?
Fundy Tech helps businesses and households across Atlantic Canada stay patched, protected and up to speed. Tell us what you’re running and we’ll tell you straight whether you need to act.