AdQuick unlocks the benefits of Out Of Home (OOH) advertising in a way no one else has. Approaching the problem with eyes to performance, created for marketers and creatives with the engineering excellence you’ve come to expect for the internet.

You can learn more at www.AdQuick.com

Hey, Josh here. This story just came out today. This is wild, scroll to the end for some of the results.

Listen, something wild happened last week that tells you everything about where we actually are with AI. Google's Nano Banana 2—their next-gen image generator—leaked online for a few hours. Not the polished, safety-filtered version they'll eventually ship to us peasants. The real one. The uncensored checkpoint their engineers play with internally.

And holy shit, the difference was stark.

Users who got their hands on it before Google yanked it down reported genuinely photorealistic outputs. Perfect prompt adherence. Lighting and textures that fooled the eye. This wasn't an incremental improvement—it was a glimpse behind the curtain at what the frontier labs have been sitting on.

Here's the kicker: this performance gap isn't an accident. It's a choice. When Google ships the official version (mid-November, with full SynthID watermarking), it'll be deliberately nerfed. Safety filters. Content restrictions. All the guardrails that supposedly keep us safe but definitely make the model dumber.

Now, I'm not saying safety measures are inherently bad. But let's be real about what's happening here: a handful of engineers at DeepMind have access to AI capabilities that would blow your mind, while you and I get the Fisher-Price version.

The leaked checkpoint—reportedly from Media.io—exposed the asymmetry we don't talk about enough. Every frontier lab has this multi-tier access system: God-mode for internal teams, progressively lobotomized versions for everyone else.

What's genuinely unsettling? The version control vulnerabilities this reveals. If an old checkpoint can leak this easily, what else is sitting in their internal repos? And as one observer noted: these engineers could cause chaos if they wanted to. That's not conspiracy theory—that's just acknowledging the concentration of power.

The uncensored model is gone now. But we saw what's possible. And we know what we're not getting.

These results are wild.

And this one:

And what it can do for throwback photos, looks like they are right from a tabloid