An interview with Sam Altman and Greg Brockman on OpenAI's restructuring, cutting Sora, “personal AGI”, Anthropic's “fear-based marketing” for Mythos, and more
The last two standing - on everything — Sam Altman and Greg Brockman came on Core Memory together for a ten-year look back at OpenAI.
Context & Ripple Effects
OpenAI’s prior coverage has framed the company around a broad product and infrastructure agenda, including ChatGPT apps, checkout, Codex research, and Sora. More recent Brockman interviews specifically put the Sora closure alongside a choice between text-focused systems and world models.
This joint Altman-Brockman interview ties that product narrowing to a restructuring and a stated “personal AGI” vision, while making Anthropic’s positioning around Mythos an explicit competitive point of contrast.
First-order effects
- OpenAI is narrowing its product portfolio by discontinuing Sora, shifting management and engineering attention away from that product while its restructuring proceeds.
- OpenAI’s leadership is publicly centering a personal-AGI narrative; Anthropic and Mythos are directly drawn into a more visible contest over how frontier AI products are framed.
Second-order effects
- The Sora decision makes OpenAI’s remaining text- and agent-oriented products, including the research direction discussed around Codex, more important tests of its product strategy.
- Competitors can use OpenAI’s withdrawal from Sora to differentiate their own multimedia or world-model efforts, while OpenAI and Anthropic are likely to compete more directly on user trust and product positioning.
Third-order effects
- If labs continue to prune costly or weakly aligned product lines, frontier-AI competition may become less about maintaining every modality and more about concentrating capital and research on a small number of platform bets.
- The emphasis on “personal AGI” raises the longer-run importance of governance for highly individualized AI systems, as product strategy increasingly depends on what users and institutions will accept as a trusted personal interface.
The trend: Frontier AI labs are consolidating sprawling research and product portfolios around a few strategic interfaces—especially personalized, agent-like systems—while competing as much on legitimacy and framing as on model capability.