Do you actually believe Mark Zuckerberg when he says AI should be accessible to everyone? Because I’ve reviewed enough AI tools with “open” slapped on the label to know that word usually means whatever the marketing team needs it to mean that quarter.
This week gave us two stories that, taken together, tell you a lot about where the AI industry’s head is at. Meta released Glimmer, an open-weight AI model, with Zuckerberg making the case that AI should be accessible to all. Meanwhile, a $250M deal involving video-clipping startup VideoVerse collapsed amid allegations of fraud. One story is about generosity, supposedly. The other is about greed, allegedly. I’d argue they’re both about the same thing: what happens when the money in AI gets so big that nobody wants to look too closely at anything.
Glimmer and the “open” question
Let’s start with Meta. Glimmer arrived in 2026 as an open-weight model, and Zuckerberg framed the release around a simple idea: AI should be accessible to everyone, not locked behind corporate gates.
Nice sentiment. But I review AI tools for a living, and I’ve learned to separate what a company says about openness from what openness actually gets you. Open weights are genuinely useful — researchers can poke at the model, developers can run it without paying API rent, and smaller shops can build on top of it without begging a trillion-dollar company for access. That matters, and I won’t pretend it doesn’t.
But “open” is not a moral position. It’s a strategy. When the biggest social media company on Earth gives away AI models, ask yourself what it gets in return. Developer mindshare. An ecosystem that standardizes on its technology. A public-relations shield that reads “we’re the good guys” every time regulators come sniffing around. None of that requires bad faith — it just requires understanding that corporations don’t do charity at this scale. If Zuckerberg genuinely believes AI belongs to everyone, great. The belief and the business advantage happen to point in exactly the same direction, which is convenient.
My test for any “open” AI release is simple: does it enable people who couldn’t otherwise afford to build? By that measure, open-weight releases like Glimmer are a real net positive, whatever the motives behind them. Just don’t confuse the product strategy with a philosophy.
Then there’s the $250M crater
Now for the other story, which is the one keeping me up at night. A quarter-billion-dollar deal involving VideoVerse fell apart amid fraud allegations. Two hundred and fifty million dollars, gone sideways.
Here’s what bothers me most. A deal of that size doesn’t happen on a handshake. There are lawyers, bankers, diligence teams, spreadsheets upon spreadsheets. And yet, allegedly, fraud made it far enough into the process that the whole thing detonated publicly. That’s not just one company’s failure. That’s a symptom of an entire sector moving too fast to check its own homework.
I see the small version of this every single week in my inbox. AI startups pitching me tools with inflated user numbers, demos that are 80% smoke, “proprietary models” that turn out to be a thin wrapper around someone else’s API. The incentive structure right now rewards the story over the substance, because the money is chasing stories. When capital floods a sector this fast, diligence becomes theater. The VideoVerse collapse is what it looks like when that theater finally gets a bad review.
Two stories, one lesson
So what connects an open-weight model release and a busted nine-figure deal? Trust, and how casually the AI industry spends it.
Zuckerberg wants you to trust that “open” means open. VideoVerse’s partners trusted that the numbers were real. In both cases, the correct posture is the same one I bring to every tool review on this site: verify, then believe. Not the other way around.
- If a company says its AI is open, check what you can actually do with it — and what you can’t.
- If a startup’s numbers look incredible, remember that “incredible” literally means “not credible.”
- If $250M can allegedly evaporate on forged foundations, your $50/month subscription decision deserves at least ten minutes of skepticism.
The AI space doesn’t have a technology problem right now. The models are genuinely impressive, and open-weight releases like Glimmer push things forward. What it has is a trust problem, and trust problems don’t get fixed by manifestos or press releases. They get fixed by people asking harder questions before signing checks — or clicking “subscribe.”
I’ll keep asking those questions. Somebody has to.
🕒 Published: