Half of Steam's new games carry an AI tag, and the tag is still too blunt

Nearly half of the games released on Steam in October 2026, 49.9 percent, disclosed that they contain AI-generated content, according to figures a Reddit user named ploz compiled and Destructoid reported.¹ That is up from 26.4 percent a year earlier and 1.7 percent in January 2024, when Steam added the disclosure option.¹ This is an opinion column, and my position is simple: the number tells us the label is working as a confession box but not yet as a buying aid.

Start with what the figure does and does not measure. The disclosure covers only content a player can see or hear, so AI used in coding never shows up.¹ Developers can also skip the tag entirely, so the real share is probably higher.¹ And the tag is one switch for very different things: a voiced character built from synthetic audio, a few placeholder textures, and a whole game stitched together from generated parts all get the same checkbox.

The spread by size is telling. Ploz found that 47 percent of games with fewer than 10 reviews carry the tag, against 14 percent of games with 1,000 or more reviews.¹ So the tag clusters in the long tail, where players already struggle to tell a careful small project from a quick one. A flag that fires on half the store and says nothing about quality is like a fuse rated for the wrong current: it trips constantly and protects nobody. Players cannot filter these games out of the store, and many were visibly irritated by obvious AI content during the recent Steam Next Fest.¹ Valve has said it is aware of the complaints but has given no timetable.¹

The best counterpoint deserves a fair hearing. AI tools can help a solo developer ship art, voice or music they could never afford, and a blanket suspicion of the tag would punish them alongside the people churning out quick-money releases. That is true, and I would not want a filter that treats every tagged game as junk. Disclosure at least gives honest developers a way to be upfront, and the rising count suggests many are using it.

But that argument is a reason to improve the label, not to leave it as is. Our earlier look at Roblox’s claim that AI tools helped publish 70,000 games points the same direction: when generation makes publishing cheap, volume rises faster than anyone’s ability to sort it. A store with a single yes-or-no tag and no filter is asking each player to open every page and judge for themselves.

What I would want is a tag that says which parts are generated (art, voice, music, text) and a player-side filter to hide or show them. Developers who use AI sparingly would stand out from those who lean on it for everything, and nobody would have to guess.

What would change my mind? If the tagged share stopped climbing and reviews showed tagged and untagged games landing in the same range, I would conclude the checkbox is fine as a plain fact sheet. I also do not know how many untagged games use AI in visible ways, because nothing in these figures measures that.

On the bench: if you buy on Steam, treat the AI tag as a starting question, not a verdict. Check the screenshots and any audio in the trailer, read a few recent reviews, and remember the tag says nothing about code.

Footnotes

1. Destructoid, “Nearly 50 percent of all of last month’s Steam releases admit to containing AI slop”. Retrieved November 2, 2026. ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷