Thoughts on the state of AI agent safety in 2026

Some notes on privacy 093 based on recent work.

Checklist

1. Define success metrics
2. Keep the first version simple
3. Add observability early
4. Document decisions

If you’ve shipped something similar, what would you do differently?