Apple foundation models limitations

Question

Level 1

6 points

Apple foundation models limitations

Apple foundation model are not accurate and NOT READY FOR RESPONSIBLE AI. how to make sure app will not have legal issues in future ?

is there a way we can have control over training and data or integration with external datasets?

secondly limitations on strict 4,096-token limit window for few simple tasks , how to meet customer expectations ?

to reproduce it just import foundation models and use playground or in the code use reasoning question and certain Health questions if you are using HealthKit integration.

MacBook Air 13″, macOS 26.4

Posted on May 14, 2026 12:29 PM

Reply

Answer 1

May 14, 2026 12:36 PM in response to bhrugu177

To be honest, there is no such thing as "Responsible AI", it is the user that is responsible for reviewing and verifying any data generated. I also would not recommend relying on Gemini results in a Google search. You may have already seen AI responses saying to eat rocks for increased mineral intake and adding Elmer's Glue to keep the cheese from sliding off of pizza.

To let Apple know about your concerns, send them your Feedback here:

Feedback - macOS - Apple

Reply