You could reply with:
I’m seeing exactly the same behavior. I’m building a journaling app, and on iOS 27 beta I’m constantly battling safety alerts that cause the model to refuse perfectly legitimate prompts. It’s become pretty exasperating. I compile the exact same app in parallel on iOS 26 devices and everything works flawlessly, so it definitely seems like something changed in the new model. Really hoping Apple tunes this.
Topic:
Machine Learning & AI
SubTopic:
Foundation Models