What happens when your free moderation tool disappears

Google is shutting down Perspective API on December 31, 2026. No migration support, and new usage or quota requests have been frozen since February. If your platform has been quietly running comment moderation through it since some engineer wired it in years ago, you have a little over three months.

The deadline itself is the least interesting part of this story. The interesting part is what it reveals about a decision most teams never actually made on purpose.

Perspective API became a default the way free things become defaults. It cost nothing, required no procurement conversation, and did something that felt like enough. A toxicity score on a 0 to 1 scale, plugged into a comment box, and the immediate problem, obviously hostile comments slipping through, looked solved. Nobody had to build a business case. Nobody had to get a budget approved. An engineer added an API call, and moderation became a line item that didn't need to be a line item.

That's not a criticism. It's exactly how a resource-constrained team should behave early on. The mistake isn't picking the free option. The mistake is never revisiting it.

The dependency nobody re-evaluates

Here's the pattern worth naming, and it isn't specific to Perspective API. Any free or nearly free vendor dependency that quietly becomes load-bearing tends to stop getting looked at. It works, so it stays off the roadmap. It's free, so it never shows up in a budget review. Nobody owns the relationship, because there was never really a relationship to own, just an API key sitting in an environment variable.

This is different from a paid vendor contract, where somebody has to sign a renewal, notice a price increase, or sit through a QBR that forces a periodic gut check on whether the tool still fits. A free dependency has no such forcing function. It just runs, silently, until the day it doesn't.

Perspective API's shutdown is a clean example because Google gave a hard date. Most of the time, the failure mode is quieter: a free tier gets restricted, a rate limit tightens, a company pivots and deprioritizes the product without ever announcing a sunset. The lesson generalizes past this one deadline.

The mistake isn't picking the free option. The mistake is never revisiting it.

What actually breaks when this happens

If you're one of the teams affected, the technical migration is the visible problem, and it's real. You need something to catch what Perspective was catching before December 31, and depending on what you built around it, that could mean re-tuning thresholds, remapping categories, and re-testing against your own traffic rather than trusting a score you never really audited in the first place.

But the more useful thing to notice is what the shutdown exposes that was already true. If losing a free API breaks your moderation, you didn't have a moderation strategy. You had a dependency you hadn't looked at closely enough to know how fragile it was.

That's worth sitting with regardless of what you migrate to next. A serious replacement, whether that's a paid API, a self-hosted model, or a full vendor relationship, is also worth periodically re-evaluating, on purpose, on a schedule, rather than waiting for the next shutdown notice to force the question.

What to actually ask before you pick whatever replaces it

Three questions worth asking about anything you're about to depend on for moderation, paid or free, that most teams skip when the current cost is zero and the current pain is low.

Who owns the roadmap, and what happens if they change direction? A free tool built by a company with other priorities can get deprioritized without warning. A paid vendor at least has a commercial incentive to tell you before something changes, but that incentive is only as strong as the contract terms actually require it to be.

What's the actual exit path if this stops working for you? The question isn't whether you can cancel, but whether you can get your historical data, your tuned thresholds, and your category mappings out in a form you can use somewhere else. A tool with no exit path is a tool you're already locked into, whether or not you've noticed yet.

Does anyone at your company actually own this relationship? The question isn't whether someone has the API key, but whether a specific person periodically checks whether this is still the right tool, still performing well, still worth the dependency. If the honest answer is nobody, that's the real finding, independent of whatever you migrate to.

If losing a free API breaks your moderation, you didn't have a moderation strategy. You had a dependency you hadn't looked at closely enough to know how fragile it was.

The actual deadline that matters

December 31, 2026 is real, and if Perspective API is running in your stack, it's worth treating as a hard date rather than something to get to eventually. Migrations that seem simple in scope have a way of taking longer than expected once you actually start re-testing against real traffic.

But the more durable takeaway isn't about this specific vendor or this specific date. It's about building the habit of asking, on some regular cadence, whether the tools sitting quietly underneath your platform's safety are still the right ones, still accountable to someone, and still something you could walk away from if you needed to. The next free dependency that quietly becomes load-bearing won't announce itself with a shutdown email either. It'll just be running, until one day it isn't.