Student Deal — 55% off annual plans. Limited time. Claim Now →
AI Detection

Content at Scale AI Detector Accuracy: What the FTC Actually Found

Content at Scale's free AI detector built its marketing around a single number — 98.3% accuracy. In 2025, the FTC investigated that claim directly, and the real figure it found on non-academic text was dramatically lower. Here's what the agency's own complaint and settlement say, not just what the marketing page claims.

Compare Against 5 Detectors Free →
✓ Free forever plan✓ No card needed✓ Used by 10,000+ students
FTC v. Workado, By the Numbers
Marketed Accuracy98.3% (claimed, academic text)
FTC-Found Accuracy53.2% (non-academic text)
OutcomeSettled FTC consent order, Aug 2025

Trusted by students at

UCLUniversity of ManchesterUniversity of EdinburghUCLAUniversity of TorontoUniversity of SydneyMcGillNYUMonashDurhamUCLUniversity of ManchesterUniversity of EdinburghUCLAUniversity of TorontoUniversity of SydneyMcGillNYUMonashDurham
1

The claim the FTC investigated

Content at Scale's AI detector is a product of Workado, LLC, and its marketing led with a specific figure — 98.3% accuracy, often rounded to "98%" across the company's site and ads. In April 2025, the FTC filed a complaint alleging that claim was unsupported, finding that the underlying detection model had actually been built by academic researchers for a narrower purpose than the company's marketing implied.

According to the FTC's investigation, the 98.3% figure came from testing against academic-style text specifically, not the general mix of writing — blog posts, marketing copy, emails — that the tool was actually being marketed to detect for everyday users.

2

What the FTC found on real-world, non-academic text

When tested on non-academic text, the FTC's investigation found the detector's real accuracy at identifying AI-generated content was 53.2% — barely above a coin flip. The FTC's Director of the Bureau of Consumer Protection, Chris Mufarrige, put it bluntly: the product "did no better than a coin toss" on that kind of text.

That's a significant gap between the headline number in the company's marketing and how the tool actually performed on the type of writing most people were using it to check.

3

The outcome: a formal, binding order

This wasn't just a critical blog post — it was a settled FTC consent order, with a final order approved in August 2025. Under its terms, Workado is barred from making AI-detection accuracy claims going forward unless it has "competent and reliable evidence" to back them up, is required to retain that evidence, has to notify customers affected by the earlier claims, and must file compliance reports with the FTC annually for four years.

No monetary penalty or consumer redress fund is part of the public order — the core remedy is the evidence and disclosure requirement going forward. Still, a formal federal order over one company's specific accuracy number is a rare, hard signal about how much to trust a self-reported percentage from any detector, not just this one.

Frequently Asked Questions

Is Content at Scale's AI detector still available?+
The FTC order didn't require the company to shut the tool down — it requires Workado to have real evidence behind any accuracy claims going forward and to notify affected customers. Whether the product itself has changed since the order is worth checking directly on the company's current site.
Does the 53.2% figure mean the tool is wrong more than half the time on everything?+
That figure specifically covers the FTC's testing on non-academic text — it isn't a universal accuracy score for every possible use case. It's still a meaningfully different number from the 98.3% the company marketed, which is the core of why the FTC acted.
Was this an FTC fine?+
No monetary penalty is part of the public consent order. It's a binding order requiring the company to substantiate future accuracy claims with real evidence, notify affected customers, and report compliance to the FTC for four years.
Are other AI detectors' accuracy claims independently verified?+
Most AI detectors, Content at Scale included before this case, self-report their own accuracy figures without an outside audit. This is one of the few instances where a regulator formally tested and challenged a specific number — worth treating any unverified vendor accuracy claim with the same caution.

Don't rely on one vendor's own accuracy number

Check your writing against Turnitin, GPTZero, Originality.ai, Copyleaks, and ZeroGPT methods at once, free, and see where they agree.

Compare Against 5 Detectors Free →
✅ Free forever✅ No card✅ Used by 10,000+ students