What GPT-5 and New AI Models Mean for Small Practices

Newer models are smarter, faster, and more contextual. But what does that mean for your risk — and your workflows? 

📦 What’s New in GPT-5 (and similar models): 

• 📚 Longer memory: Better follow-through across a conversation or note 

• 🎯 Fewer hallucinations: More accurate clinical output — if prompted well 

• 🧠 Faster comprehension of medical context (with better pretraining) 

 🩺 Risks and Opportunities: 

✅ Opportunity: 

• Faster SOAP note drafting 

• Better phone triage summaries 

• Smarter intake prep (customized scripts, reminders) 

❗ Risk: 

• Overtrusting outputs: Providers are more likely to copy/paste without review 

• Deep contextual data = higher PHI risk 

• More data “leakage” if APIs aren’t secure or logged 

 ✅ What You Should Do: 

• Use HIPAA-safe interfaces only (e.g., Azure OpenAI with logging) 

• Build review steps into any automation 

• Never fully automate diagnosis, consent, or decision-making without human oversight 

• Choose models with audit and deletion tools 

 🧭 Pro tip: 

The smarter the model, the stronger your policies must be. Invest in prompt templates, risk logs, and AI SOPs now to avoid problems later.