Third-party cyber evaluations involving OpenAI models
openai.com·13h ago
Data Science
Simon Willison's Blog·Jun 8·5 min read
Siri AI at WWDC 2026
TL;DR
Developers faced challenges integrating AI features into existing applications without custom code. Apple introduced a custom Gemini-derived model and a Core AI library that integrates with Meta's PyTorch ecosystem.
✦ Why It Matters
Engineers can now leverage Apple's hardware for AI model deployment without extensive custom coding.
Key Takeaways
- Explore the new Core AI library for enhanced model performance on Apple devices.
- Utilize vision LLMs to improve user interaction without custom app code.
- Check the iOS 27 Developer Beta for early access to Siri AI features.
- Stay updated on user reports from those who accessed the new Siri AI.
How It Works
The new Siri AI features leverage vision LLMs to analyze and extract information from the user's screen, allowing for more intuitive interactions without requiring developers to modify existing applications. The Core AI library integrates with PyTorch, enabling developers to convert their models for optimized performance on Apple hardware.