| 1 | [Transition] So where does this leave us? |
| 2 | |
| 3 | Auto Mode catches enough dangerous actions to make autonomous operation substantially safer than running with no guardrails. The classifier doesn't need to be perfect to be valuable. |
| 4 | |
| 5 | We'll continue expanding our overeagerness testset, iterating on both safety and cost. We encourage users to stay aware of residual risk, use judgment about which tasks to run autonomously, and — importantly — tell us when Auto Mode gets things wrong. |
| 6 | |
| 7 | [Pause] |
| 8 | |
| 9 | You can start right now with claude auto-mode defaults. Thank you. |
| 10 | |
| 11 | Key points: ① Substantially safer than no guardrails ② Not a replacement for careful review on high-stakes tasks ③ Available now — try it |
| 12 | Duration: 1 minute |