The FTC’s DoNotPay order required $193,000 in monetary relief, notices to certain past subscribers and an end to unsupported claims that the service could substitute for a professional. The case was about alleged deceptive marketing and a lack of evidence for claims of legal expertise—not a ban on AI. Its broader lesson for AI companies is straightforward: substantiate the specific promises you make about performance, safety, accuracy or financial results, and keep that evidence current.
What the FTC’s DoNotPay order required
The FTC alleged that DoNotPay marketed its service as “the world’s first robot lawyer” and represented that it could perform like a human lawyer. The complaint described claims that the service could apply law to a user’s particular facts, account for legal complications, generate legally valid documents, identify legal violations on small-business websites and help consumers pursue claims without a lawyer. The agency alleged those representations were false, misleading or unsubstantiated when made. Read the FTC complaint.
The FTC said DoNotPay had not tested whether its chatbot’s output was equivalent to a human lawyer’s work and had not retained attorneys to validate the accuracy and quality of law-related features. In January 2025, the FTC finalized an order publicized on February 11 that imposed $193,000 in monetary relief, required notices to consumers who subscribed between 2021 and 2023, and prohibited unsupported claims that the service could substitute for a professional service. The order resolved the FTC’s allegations; it was not a ruling that AI legal assistance is inherently unlawful. See the FTC’s announcement of the final order.
The distinction matters to companies and consumers alike. A tool that offers legal information, helps draft a first-pass document or supports lawyers is not the same as a service marketed as an autonomous legal representative. The closer a product’s promise comes to professional substitution or guaranteed outcomes, the more its claims need to be supported by evidence matching that promise.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →#1 Best Overall
- Built for Local AI Development: AMD Ryzen AI Halo is designed for local AI development and inference, featuring 128GB unified memory and support for up to 200B parameter models to build and run intensive AI workloads locally.
- 128GB Unified Memory: Features 128GB LPDDR5x unified memory at 8000 MT/s with 256 GB/s memory bandwidth, providing a shared memory pool across the CPU, GPU, and NPU to support larger AI models.
- AMD Ryzen AI Max+ 395 Processor: Features 16 cores, 32 threads, and Zen 5 architecture, paired with AMD Radeon 8060S integrated graphics featuring 40 RDNA 3.5 compute units and an AMD XDNA 2 NPU with up to 50 TOPS.
- Linux AI Developer Platform: Purpose-built for Linux-based AI development with full AMD ROCm software support and preloaded tools, models, and workflows optimized for local AI development.
- Compact, Connected Design: Includes a 2TB M.2 SSD, 10GbE LAN, Wi-Fi 7, Bluetooth 5.4, USB-C connectivity, and HDMI 2.1b.
Why the case reaches beyond legal technology
The FTC’s concern was the gap between what a product can sometimes do and what its marketing leads customers to believe it can reliably do. A model’s fluent answer, a successful demo or a benchmark result does not automatically establish that a complete commercial product delivers the advertised result in real customer workflows.
That principle applies to claims about accuracy, speed, safety, detection, cost savings, revenue, automation and professional equivalence. For example, “helps flag possible issues for review” is narrower than “catches every error”; “can assist with a defined workflow” is different from “replaces an employee.” Softer wording is not automatically compliant: the overall impression must still be truthful and not misleading.
Use of a third-party model does not automatically transfer responsibility for a company’s own customer-facing promises. Vendor contracts may allocate costs or indemnity between businesses, but they do not necessarily prevent regulators from scrutinizing the company that markets the finished service.
What the FTC’s wider AI actions show
In September 2024, the FTC announced Operation AI Comply, an enforcement sweep applying existing consumer-protection principles to alleged deception and unfair conduct involving AI. The matters involved different theories, not one uniform AI violation or a new comprehensive AI statute. The FTC’s announcement describes the cases.
- DoNotPay: Alleged unsupported claims that an AI service could substitute for a human lawyer.
- Automators / FBA Machine: Alleged AI-powered business-opportunity and earnings claims.
- Career Step: Alleged deceptive career-training and employment claims involving AI-related representations.
- NGL Labs: Alleged claims about AI moderation in an anonymous messaging app marketed to children.
- Rite Aid: Alleged unreasonable safeguards around facial-recognition technology.
- CRI Genetics: Alleged deception concerning DNA-report accuracy and AI-based genetic matching.
Later actions reinforce the range of potential issues. In April 2025, the FTC announced an order involving Workado and claims that its AI-detection product was 98% accurate; the agency said effectiveness claims require competent and reliable evidence. That figure is an allegation-specific marketing claim, not a general finding about AI-detection products. A percentage claim should identify what was tested, the dataset and conditions, and error rates that matter to users. Read the FTC’s Workado announcement.
Rank #2
- EVOLUTION AMD RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 64GB pool, which is perfect for running LLMs such as Deepseek 32B, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 4% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
The FTC’s AI enforcement index also identifies action involving Rytr and services dedicated to generating consumer reviews or testimonials. That illustrates a distinct risk: a product can create exposure through what it enables customers to publish, especially fabricated reviews or endorsements. See the FTC’s AI enforcement page.
In March 2026, the FTC announced a proposed settlement with Air AI over alleged business-growth, earnings-potential and refund-guarantee claims. The proposed $18 million monetary judgment was largely suspended based on inability to pay, and the proposed settlement included a ban on marketing business opportunities. The allegations concern a different kind of claim from DoNotPay’s professional-equivalence claims, but show that AI vendors selling to businesses can also face scrutiny over promises of revenue, passive income, guaranteed results or cost elimination. Read the FTC’s Air AI announcement.
How to build evidence for an AI performance claim
Evidence should match the exact claim, product and context—not merely the underlying model in isolation. A useful evaluation is planned before the claim is published, uses representative conditions, measures meaningful failures and can be reproduced from retained records. The stakes matter: creative-writing assistance and systems touching legal rights, health, finances, employment, housing, children or safety do not call for the same level of scrutiny.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →- Write down the claim precisely. Record the exact public wording and the reasonable impression it creates. Separate factual descriptions from promises about performance, savings, earnings, safety or replacement of a professional.
- Define the tested product and use case. Identify the production model and version, prompts, retrieval sources, interface, workflow, intended users, languages and jurisdictions. A model benchmark alone does not establish how the full service performs.
- Design a realistic evaluation. Use a sufficiently representative test set and conditions that reflect actual use. Measure false positives and false negatives where relevant, along with failure modes and the consequences of errors. State sample size, methodology and uncertainty rather than presenting a bare accuracy percentage.
- Set a risk-appropriate review process. For consequential use, provide a meaningful route to qualified human review. Human review is not a cure if reviewers lack time, expertise or information to catch errors.
- Keep a claim evidence file. Retain the wording, test dates, version details, data provenance, sample size, methods, results, limitations and any independent review. Keep approvals from responsible product, legal or compliance staff where appropriate.
- Re-test changes and monitor real use. Version evaluations, run regression tests after model, prompt, retrieval or interface changes, track complaints and error reports, and revise or withdraw claims when performance no longer supports them.
For a particular accuracy claim, explain what the metric measures and under which conditions. A result on a benchmark does not prove equivalent performance across different populations, languages, jurisdictions or customer workflows. Claims such as “better than humans” need especially careful definition: better at which task, compared with whom, measured when and with what error costs?
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Why disclaimers and beta labels do not replace proof
Clear disclosures help users understand a product’s limitations, but small-print language may not cure a prominent, contradictory claim. “Results may vary,” “for informational purposes only,” “not legal advice” or “AI can make mistakes” does not itself substantiate a promise that a service replaces a lawyer, guarantees accuracy or prevents losses.
Rank #3
- Intel Core Ultra 9 285 Processor: Newly developed cores deliver ultra-smooth and responsive gameplay. AI accelerators prepare users for the next era of gaming on an AI PC.
- Simplistic Design: Enjoy the latest generation of Windows 11 Home for your everyday needs. *MSI recommends Windows 11 Pro for business use.
- NVIDIA GeForce RTX 5070 Ti GPU
- Cool While Gaming: In conjunction with an RGB CPU Air Cooler, the Aegis RS features four system cooling fans; three in the front and one in the rear to pull in cool air and push heat out of the PC.
- Turn on the Bright Lights: With the built-in RGB lighting, take your gaming experience to the next level by pressing the MSI LED button to cycle through lighting options. Customize lighting even further with MSI Center software.
Likewise, “beta,” “experimental” or “early access” labels can set expectations but do not license production-level promises unsupported by evidence. Terms of service and a probabilistic architecture are not automatic defenses to misleading advertising. Disclose limitations where they matter to a purchasing or usage decision, and design the product so its actual use aligns with the claim.
A practical risk framework for AI claims
The following tiers are a planning aid, not a legal classification. The closer an error could come to harming a person or changing an important outcome, the more rigorous testing, safeguards and review should be.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errors- Lower consequence: Brainstorming, formatting and creative assistance. Evaluate quality and known limitations, and avoid implying factual reliability beyond what is tested.
- Moderate consequence: Business workflow automation, customer support and document analysis. Test representative inputs, escalation paths, privacy risks and the effect of errors or rework.
- High consequence: Legal, medical, financial, employment, housing, child-directed, biometric or safety-related uses. Assess the applicable sector and privacy rules, discrimination and safety risks, and whether qualified human review is needed.
- Highest-claim risk: Promises of professional substitution, guaranteed outcomes, universal detection or earnings. These claims require especially strong, claim-specific substantiation; narrowing or removing the promise may be more defensible than trying to qualify it with fine print.
A constrained system—limited to a defined workflow, jurisdiction, knowledge base or document set—is generally easier to evaluate than an open-ended assistant marketed across many domains. Narrowing intended uses can make both the product and its claims clearer, though it does not by itself establish compliance.
What AI companies should do now
- Inventory every claim. Review the homepage, landing pages, app-store descriptions, social posts, videos, sales decks, affiliate copy, customer success stories and other public or private marketing—not just the current homepage.
- Map claims to evidence. Create a register linking each material claim to its test, version, limitations and approval. Pause or revise claims for which evidence is missing or no longer representative.
- Check customer-enabled claims. Do not supply templates or incentives that encourage unsupported testimonials, fake reviews or exaggerated customer results. Review published case studies and endorsements for substantiation.
- Build cross-functional ownership. Product, engineering, security, legal, marketing, sales and customer support all affect what users are promised and what failures are detected.
- Plan for other applicable rules. Depending on the product, consider privacy and data security, children’s privacy, professional licensing, state consumer-protection rules, employment or housing discrimination, financial-services requirements, intellectual-property issues, and rules for reviews and endorsements. The DoNotPay order itself does not create a comprehensive AI regulatory regime.
- Prepare a challenge response. If a claim is questioned, preserve relevant records, assess potential customer harm and refund obligations, re-test realistically, correct public statements promptly and consider notifying affected users. Do not quietly change a product while leaving outdated marketing in place.
What the DoNotPay case does not mean
- It does not categorically prohibit AI legal-information, drafting or lawyer-support tools.
- It does not establish that every AI output must be perfect.
- It does not make a disclaimer a substitute for truthful marketing and supporting evidence.
- It does not impose one identical test or review process on every AI company.
The FTC’s case page contains the complaint, order and related filings for readers who want the procedural record. View the DoNotPay case materials.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
