“I personally do not think AI will kill all humans”
Barak directly responds to Hubinger by expressing respect for researchers at Anthropic, calling for coordination, and saying he does not personally expect AI to kill every human. He nevertheless warns of multiple bad outcomes if safety is not prioritized.
- Speaker
- Boaz Barak
- Capacity
- Personal public post by Boaz Barak, speaking about OpenAI from inside the organization; not presented as an official OpenAI statement.
- Observed
- September 10, 2026
- Published here
- September 10, 2026
- Classification rationale
- Questions universal-extinction expectations while supporting safety prioritization and coordination; does not disprove Hubinger's probability.
Surrounding material
Full context
Before the quotation
The retrieved primary thread includes Hubinger's parent post stating a personal extinction-risk estimate above 10% within a decade and conceding that Anthropic lacks a solved alignment plan for superintelligence.
After the quotation
Barak's same-period separate post says OpenAI's awareness of the magnitude of AI capabilities has changed recently and acknowledges both potential benefits and many ways development could end badly. His reply gives no numerical risk estimate.
Evidence
Complete source record
Full primary public post and available surrounding thread. Related context: https://x.com/boazbaraktcs/status/2097514504728748467 https://x.com/EvanHub/status/2097497037956891126 Speaker and position scope are explained in the linked event.