Stability AI

This wasn't a revelation from him, nor was it the first time actually that models had done something like that.

The tweet has since been deleted, but you can imagine that some people, like the former founder of Stability AI, felt like this was a bit of policing gone too far.

You can imagine some developers nervously not using Claude 4 Opus, thinking it might call the cops.

In the clarifying tweet, Sam Bowman confirmed that this isn't a new Claude feature, and it's not possible in normal usage.

They've got Claude Three Five Sonnet from Anthropic, which is genuinely competitive with GPT-4.

They've got Stability AI's image generation models.

They've got specialized models from AI21 Labs and Cohere for specific enterprise tasks, and they've got their own Titan models that are optimized for things like embeddings and code generation.

But here's the thing nobody's talking about.

But first, let's talk about what just happened that got everyone in a tizzy.

So we've been hearing about AI for decades now.

[whimsical music] [objects clattering] [person laughing] It's always been just around the corner, but in the last two years or so, it went from someday, maybe, to right here, right now, deal with it.

AI art is getting first places at art competitions.

Professors can't distinguish AI-written essays from human-written ones.

Scientists are tricked by AI-written scientific abstracts.

We're hitting this inflection point where AI progress went from slow, boring, steady to rapidly picking up speed, and no one is ready.

It's gonna be chaos, and you know what chaos is.

So we're going to talk about what's unfolding there.

We want to talk about AI, or as President Trump calls it now, SI, and how that is just growing at a rate that we just can't even keep up.

You'd wait to hear what they've done.

It's unreal.

So we'll be talking about that as well.

There's been some breaches of confidentiality with the FBI and OpenAI there, so we'll see if we touch on that.

And then we want to go to Ilhan Omar.

Bit of an interesting story unfolding there.

Industry dynamics complicate things further.

Adobe, Stability AI, and other major players have significant investments in current personalization and motion capture approaches.

They're not gonna simply abandon their existing technology stacks because NVIDIA published impressive papers.

Market adoption depends on integration partnerships, and those negotiations can take years while introducing compromises that degrade the original research quality.