The ‘AI Ported Age of Empires II’ Demo Is Lying to You

You’ve probably seen the video by now. Codex — an AI coding tool — allegedly ported Age of Empires II to run in a browser. The tweet went viral. The comments lit up. And somewhere in the back of your mind, a quiet voice whispered: Wait, where’s the link to play it?

That voice was right.

Here’s the thing about magic tricks: they only work when you can’t inspect the apparatus. The moment the magician invites you backstage, the illusion collapses. And right now, nobody — not the poster, not the commenters, not you — can find a single playable link or a single line of source code.

A demo video without a working product isn’t a breakthrough. It’s a commercial.

Let’s be clear about what we’re looking at. Porting a game like Age of Empires II to a browser isn’t just translating code from one language to another. It’s handling thousands of edge cases — pathfinding bugs, multiplayer sync, save file compatibility, rendering quirks across browsers, audio latency, memory management. The video shows you the happy path. The happy path is the easiest thing in the world to fake.

You know what’s harder than showing AI write a port? Showing AI write a port that actually works when a million players click “start game” simultaneously.

The top comment on the original tweet cuts through the hype with surgical precision: Seems to be video only, no link to source nor to try it. That’s not cynicism. That’s pattern recognition. We’ve been here before — the AI demo that looks too good to be true, the cherry-picked output that crumbles under real-world conditions, the viral claim that quietly disappears when someone asks for proof.

If you follow AI development, you know the cycle by now:

1. Someone posts an impressive demo.
2. The internet loses its collective mind.
3. Engineers ask for source code.
4. Crickets.
5. Everyone moves on to the next shiny thing.

We’ve trained ourselves to be amazed by AI and unlearned the habit of asking ‘but does it actually work?’

Here’s what the demo deliberately doesn’t show you: what happens when the AI encounters a legacy code pattern it wasn’t trained on. What happens when the ported version hits a race condition the original game never had. What happens when you try to load a save file from 2003. These are the edge cases that separate a real port from a simulation of one.

And that’s the real story here — not whether Codex can port a game, but how we’ve collectively lowered the bar for what counts as proof. A video. A tweet. No code. No link. No scrutiny. Just vibes.

The most dangerous thing about AI isn’t what it can do. It’s what we’re willing to believe it can do without evidence.

If this were a real port, the source code would be on GitHub by now. The playable link would be in the bio. The developer would be fielding bug reports instead of quote-tweets. Instead, we get a polished video and a thread full of people asking the same question: Can I try it?

Silence.

That silence tells you everything.

A single video proves nothing. The absence of verifiable proof is itself the proof that something’s off.

AI’s ability to port software will be measured not by what it shows on camera, but by the robustness of the code it produces across every edge case, every failure mode, every unglamorous scenario that doesn’t make for good social media content.

So the next time you see an AI demo go viral, ask yourself one question: Where’s the link?

If there’s no link, there’s no product. And if there’s no product, you’re watching an ad.