What This Actually Cost Real People

Picture a college junior in 2023, a first-generation student who spent three nights rewriting a personal essay. She submits it to her professor, who runs it through an AI detector. The tool highlights half her sentences in red and orange. The platform’s interface tells her, with the authority of a number, that her work “reads very robotic and therefore has a higher chance of being AI generated.” Her professor opens a disciplinary hearing.

She didn’t use AI. The detector was broken. And the company selling it knew, or should have known, that it was broken.

This is the injury that doesn’t show up in a settlement spreadsheet. There is no dollar figure for the humiliation of defending your own writing to an institution that trusted a $49-per-month subscription tool over your word. There is no line item for the anxiety that follows a student through the rest of her academic career, always second-guessing whether her own voice sounds “human enough” to survive a machine’s judgment.

For journalists, the stakes were professional and immediate. A freelance writer whose pitch gets flagged by an editor running a Content At Scale check doesn’t always get to appeal. Editors are busy. Flagged content gets cut. Relationships erode. The byline goes to someone else. The writer never learns why.

For marketers, there was a specific cruelty to this particular scam. These were people who paid for the tool precisely because they needed to know whether their AI-assisted copy would pass detection elsewhere. They were told, with precision down to a tenth of a percent, that the tool would tell them the truth. Instead, they were handed a coin and told it was a compass.

The rebranding from “Content At Scale AI” to “Workado, LLC” didn’t happen in a vacuum. Corporate name changes after regulatory scrutiny are how companies try to walk away from their track record and start over with a clean slate. The name changes. The conduct stays the same. The harm is already done.

What They Said. What the FTC Proved. Word for Word.

These are direct quotes from the FTC complaint and from the company’s own captured advertisements. Nothing is paraphrased. Nothing is invented.

“Our AI Detector can predict with 98.3% accuracy whether your content is human or AI-generated. It checks for traces of AI in written text from sources like ChatGPT, Claude, Bard, GPT-4, and more.” — Content At Scale advertisement, captured October 3, 2023 (Exhibit A, FTC Complaint)

“Trained on blog posts, Wikipedia, essays, and more. This AI Checker looks for patterns that indicate AI-written text (such as repetitive words, lack of natural flow, and generic tone), forecasts probable word choices, and analyzes sentence structure and other characteristics for 98% accuracy.” — Content At Scale advertisement, captured October 3, 2023 (Exhibit A, FTC Complaint)

“In fact, it’s so good, it will even flag human-written text as AI if it sounds robotic. (Let’s face it not all of us are great writers. In this case, the AI Detector will help you write more naturally, with better readability!)” — Content At Scale website copy, captured October 3, 2023 (Exhibit A, FTC Complaint)

“Respondent did not build, train, or fine tune the AI model behind its AI Content Detector. The AI model behind Respondent’s AI Content Detector is publicly available on the Internet… The developers of this publicly available model were students in Norway who developed it for an undergraduate thesis.” — FTC Complaint, Paragraph 10

“The developers’ best result for the AI Model’s accuracy when evaluating a mix of human-created and AI-generated non-academic content was 74.5%. The developers’ testing data also showed that the AI Model struggled to identify AI-generated content as AI-generated when evaluating non-academic content, correctly detecting AI-generated text merely 53.2% of the time.” — FTC Complaint, Paragraph 14

Who Got Hurt and How

Public Health of Truth and Trust

A tool that claims to distinguish human from machine communication, and fails at it, poisons the information ecosystems it enters. The FTC complaint identifies several specific harm pathways.

• Students who write genuine work can be accused of academic dishonesty when their writing style is flagged by a tool that was not designed or tested for educational text. The complaint explicitly names this scenario: “a student wrongly accused of cheating” is the FTC’s own framing of the foreseeable harm.
• Journalists submitting work to editors who use AI detection tools can have articles rejected based on false positives. The complaint names this specifically: “a journalist’s article rejected for publication.” This damages both individual careers and the broader supply of human-generated reporting.
• The tool actively told users that even human writing would be flagged as AI if it “sounds robotic,” normalizing the idea that the tool’s errors were the writer’s fault. This shifted accountability from the product to the person harmed by the product.
• By claiming to detect content from Claude, Bard, and GPT-4 when it had never been trained on any of them, the tool created false confidence in institutions relying on it to enforce academic integrity policies, potentially leading to systematic misjudgment of student work across multiple platforms.

Economic Inequality

The damage from this tool fell hardest on people who could least afford it: workers in low-margin creative economies and students without institutional protection.

• Freelance writers and marketers paid $49 per month for a tool they believed gave them a reliable signal about whether their work would pass AI detection by clients and publishers. They were paying for certainty and receiving a coin flip.
• The “pro version” with “suggested rewrites of content flagged as AI-generated” meant customers were also paying for a second product to fix the false positives generated by the first. The error was monetized twice.
• Students, who are typically without legal resources, faced institutional discipline proceedings based on tool outputs. Contesting an academic dishonesty finding requires time, knowledge of appeals processes, and often legal or administrative support that low-income and first-generation students systematically lack.
• The company sold subscriptions from September 2023 through August 2024, a full eleven months, after the $49/month tier was added. Every dollar collected during that period was collected on the basis of accuracy claims the FTC found to be false or unsubstantiated.
• Content creators in the marketing industry, already operating in a compressed economy where AI tools drive down rates, were told this product would help them stay competitive. A tool with a 53.2% detection rate on the content they actually create does not help anyone stay competitive. It generates noise.

What They Charged. What It Was Worth.

Who Is Responsible. What Can Be Done.

The FTC has filed its complaint. The following corporate roles and regulatory bodies are now part of this public record. Named executives are not identified in the source document; the complaint names the entity only.

Corporate Accountability

• Workado, LLC (formerly Content At Scale AI): Principal respondent. Registered in Arizona. Business address on record: 15333 N. Pima Road, Suite 260, Scottsdale, Arizona 85260.
• Leadership of Workado, LLC: Names of individual officers and board members are not identified in the source complaint. Public incorporation records for Arizona LLCs are accessible through the Arizona Corporation Commission and may identify registered agents and managing members.
• Tool still accessible: As of the complaint, the AI Content Detector remained available at contentatscale.ai and brandwell.ai domains. Anyone currently using these tools should treat their outputs as unreliable for any consequential purpose.

Watchlist: Regulatory Bodies

• Federal Trade Commission (FTC): Filed the complaint under Docket No. 232-3092. The FTC’s consumer protection division handles AI tool deception claims. File complaints at ftc.gov/complaint.
• Consumer Financial Protection Bureau (CFPB): Relevant if subscription billing practices involved deceptive renewal or cancellation terms not addressed in this complaint.
• State Attorneys General: Arizona AG has jurisdiction over the registered entity. Additional states’ AGs have standing where consumers were harmed within their jurisdictions.
• Department of Education: Has regulatory interest in AI detection tools used to adjudicate student academic integrity. No federal guidance currently mandates disclosure of detection tool accuracy rates to students.

Mutual Aid, Organizing, and Resistance

• If you were accused of AI-generated content using this tool: Document the accusation, the tool used, and the date. The FTC complaint establishes a public record of the tool’s inaccuracy that can be cited in institutional appeals. Academic appeals offices are required to consider exculpatory evidence.
• If you paid for the $49/month subscription: File a complaint with the FTC at reportfraud.ftc.gov. Include dates of subscription, any documented harm (content rejected, account flagged, work accused of being AI-generated), and screenshots if available.
• If you are an educator using AI detection tools: This case is a documented reason to require that any AI detection tool used in academic discipline proceedings disclose its trained domain, tested accuracy rate, and false positive rate on human text before results are used against a student.
• For writers and content workers: Organize through freelance and journalist unions to establish workplace standards requiring that editors and publishers disclose which AI detection tools they use and what their documented accuracy rates are. The National Writers Union and the Freelancers Union are entry points.
• For developers and researchers: The model at the center of this case, the RoBERTa-academic-detector by andreas122001, is publicly documented. Publishing clear user-facing warnings about its narrow training domain directly on the repository page would limit further misappropriation.