[furyofantares]

That is my best guess too, which is an even better case for better-than-Opus-4.8 models being available from OpenAI without drawing the attention of the administration.

I do think their marketing hype was a factor though even in this narrative. Andrew Jassy tells a vengeful admin that there's this jailbreak for a model Anthropic said was dangerous, and vengeful admin seizes the opportunity to take vengeance.

[ethbr1]

I think any optimism about future OpenAI models is discounting the avarice of the current US administration.

Having succeeded with Anthropic, it's reasonable to expect their next attempt will be:

"Nice new model, OpenAI, would be a shame if national security concerns tangled up a public release. Now let's talk about equity stakes..."

[furyofantares]

My belief that OpenAI will release a model better than Opus 4.8 globally should not be interpreted as a claim that they will avoid a shakedown. I think if they encounter corruption, they will engage with it.