Workado AI Detector: Accuracy Claim Evidence Questions

What was claimed

Workado's AI content detector was marketed as capable of identifying AI-generated text with a stated accuracy figure — framed as a reliable indicator of content origin without qualification on content types, model versions, or test conditions.

Risk pattern

Numeric accuracy without disclosed test scope or error rates

Why this mattered

A numeric accuracy figure for an AI detector is only meaningful if the test design, content types, AI model versions covered, sample size, and error rates are disclosed. Without these details, buyers cannot evaluate whether the stated accuracy applies to their use case. Claims that omit this context leave the evidence burden unmet.

What the source said

The FTC finalized an order finding Workado misrepresented the accuracy of its AI content detector. The order required Workado to retain evidence substantiating any future accuracy claims and prohibited unsubstantiated accuracy representations.

Evidence gap / buyer questions

Benchmark design and scope, content types included in testing, AI model versions covered, sample size, false positive rate, false negative rate, and how frequently the benchmark is updated as AI models change.

• What content types and AI model versions were included in the accuracy test — and is the test set documented?
• What is the false positive rate for content this detector incorrectly labels as AI-generated?
• How often is the benchmark updated as new AI models are released?
• Is the detector output intended as a final determination, or as a signal requiring human review?

Wording boundary direction

On [named benchmark] using [described content types] against [AI model versions] as of [date], the detector reported X% match with ground-truth labels. False positive and false negative rates are described at [link].

A lower-risk wording boundary narrows the scope, discloses the test conditions, and does not overstate what is covered.

Update and response status