MarqVision
MarqVision is an AI brand protection platform that finds and removes infringements of a brand's intellectual property online. Its detection engine scans more than 1,500 marketplaces, social platforms, websites and app stores across 118 countries, comparing what it finds against the brand's own product data at SKU level so that it catches fakes which copy specifications and imagery while avoiding the logo, and it analyses video frame by frame to catch counterfeits pushed through social clips. Around that sit four further lines: impersonation, covering lookalike domains, scam advertisements and spoofed social profiles, where a multi-agent system decides whether an infringement is valid and classifies the grounds as phishing, counterfeit or trademark abuse, returning its reasoning and a risk score with each classification; content protection for piracy and illegal listings, including direct takedown authority on YouTube; unauthorised sales, tracking grey-market resellers and pricing hourly across a database of more than ten million sellers; and MarqLaw, which coordinates customs work, criminal enforcement, field investigations and litigation through partner law firms in the United States, Korea, China and Southeast Asia. A separate product, MarqFolio, handles trademark registration, renewal and monitoring. Enforcement runs on a letter of authorisation from the brand, with a customer-maintained whitelist of sellers it does not wish to act against, and the company describes a tiered model in which borderline cases route to human analysts and a reviewer confirms significant decisions. Customers include Nissan, GM, Panasonic, Henkel, Miele, Lush, Wella, Coupang, Shinsegae and MSCHF. MarqVision is operated by Marq Vision Inc., a Delaware corporation with offices in Los Angeles, San Francisco, New York, Seoul, Paris, Shanghai and Tokyo.
Capability grades
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
AI Centrality
How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.
The models are the product and there is nothing conventional beneath them. What MarqVision sells is detection and enforcement at a scale no human team can reach: scanning more than 1,500 channels across 118 countries, comparing findings against a brand's own product data at SKU level, analysing video frame by frame, and generating the notices that follow. The named capabilities are all machine learning: Atomic Product Detections described as using generative AI for SKU-level precision, a multi-agent system that decides whether an infringement is valid and classifies the enforcement grounds, an LLM framework that auto-generates search keywords in any language, image recognition and text analysis, and models the company states are trained on successful enforcement data. There is no content asset underneath, since the corpus is marketplace listings the crawler gathers and only the models make usable, and the stated comparator is not other software but the manual monitoring the product replaces. Verified 2 September 2026.
Citation Accuracy and Hallucination Disclosure
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
Accuracy is measured, published, and unusually well contextualised, and one clause in the agreement undercuts all of it. Three figures appear across the product lines: 97 per cent on enforceable item detection using SKU-level comparison, 99.8 per cent across more than 48,000 domain impersonation incidents, and 94 per cent SKU-matching accuracy on unauthorised sales, with the company stating openly that these are internal benchmarks. The mechanism behind them is described rather than asserted, covering SKU-level comparison against genuine product data, multi-signal detection and a tiered model routing borderline cases to analysts. Failure modes are named in more detail than any vendor in this corpus manages, in a published piece that sets out what a false positive costs and names misrepresentation liability under 17 U.S.C. 512(f), tortious interference and platform reporting suspension as the consequences. What holds this below the top band is that only one figure carries a sample size and none carries a test set, and more importantly that section 2.4 of the operative terms states that Marq Vision assumes no liability for the quality, accuracy or validity of the data gathered in or by the Platform and that in no event shall the customer rely on it. The marketing measures accuracy and the agreement disclaims reliance on it. Verified 2 September 2026.
Autonomy and Oversight Model
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
Real review surfaces are published and the threshold at which the system acts alone is not. On the oversight side the material is substantive: the company states that a human reviewer confirms every significant decision, describes a tiered human-in-the-loop model that routes only borderline cases to analysts, and returns the reasoning and a risk score with each classification so a team can see why a determination was made. Two customer-side controls are contractual rather than optional. Section 8.3 of the terms requires the customer to have qualified representatives using due care to review, validate and verify all information transmitted, and to report claims only where it holds a good faith belief in their veracity. The whitelist gives the customer a standing exclusion list of sellers it does not wish to enforce against. Against that sits the speed the product is sold on, with 15-minute detection-to-enforcement, median responses under five hours for fake domains, response times described as seconds for piracy, and enforcement actions initiated automatically. Both statements cannot be fully true at once, and what the conflict shows to be missing is the definition: significant and borderline are never defined, so nothing published states which determinations reach a person. Verified 2 September 2026.
Operational and Outcome Evidence
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
The reference base is one of the largest in the index and spans categories that rarely appear together. Roughly forty logos run across automotive with Nissan and GM, electronics with Panasonic, consumer goods with Henkel, Persil and Miele, beauty and fashion with Lush, Wella, Stussy, Courreges, Lemaire and Ader Error, Korean commerce and finance with Coupang, Shinsegae, Toss and KT&G, and gaming and media with Smilegate and Class101. Six case studies are published and five individuals speak with name and title, including Tae Hyun Kim, Director of Ecommerce at Miele, Ryan Dahlstrom, Global Director of Digital Commerce at Darn Tough Vermont, and John Belcaster, General Counsel of MSCHF. Operational figures are published in quantity: 180 times faster enforcement for its own teams, 50 per cent faster than traditional providers, 15-minute mean time to resolution, more than ten million sellers tracked, dozens of offline raids in Shenzhen, Guangdong and Yiwu, and one million dollars of counterfeits seized. What holds this below A is method and dating: no figure carries a measurement basis or a date, and none is tied to a named customer. Verified 2 September 2026.
Privilege and Confidentiality Posture
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
One limb is genuinely strong and the rest are thin or point the wrong way. Section 5 of the terms is a proper mutual confidentiality regime with a compelled-disclosure notice commitment, and it survives termination. Section 4.2 confirms the customer owns all data it transmits and that no licence is granted beyond providing the services, with section 4.3 limiting Marq Vision's licence to the term and to service provision. Against that, section 2.5 grants a broad right to use customer-submitted information, expressly including information about the customer's customers, distributors, inventory and products, for improving the platform, developing updates and improvements, and improving Marq Vision's, its subsidiaries', its vendors' or its other customers' business models and services. Anonymised data is carved out of confidential information entirely at section 5(d) and is owned by Marq Vision under section 4.1. Nothing addresses segregation between customers, no retention or deletion position for customer data appears in either the terms or the privacy policy beyond a general reasonableness statement, and neither privilege nor work product is mentioned, which is defensible given section 1.2 disclaims any attorney-client relationship but leaves the question unanswered for a customer whose in-house counsel routes enforcement through the platform. Verified 2 September 2026.
UPL and Professional Responsibility Posture
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point.
The agreement draws the line clearly and the marketing crosses back over it. Section 1.2 of the terms, headed Authorization; No Legal Representation, states that the services do not include any legal representation, that Marq Vision and its personnel do not and cannot provide legal guidance or advice, that in no event shall the services constitute the creation of any potential or actual attorney-client relationship, that soft notices are not a substitute or replacement for any legal claims, and that any legal questions should be directed to the customer's own counsel. It also states that where offline services involve proceedings, those are brought by third-party law firms or agencies. That is a complete statement of what the product is and is not, and it is reinforced by the customer warranties at section 8.3 requiring due care in review and a good faith belief before any claim is reported. What pulls against it is a marketing claim on the anti-counterfeit page that the AI-driven solutions comply fully with the ABA's ethics guidelines, ensuring the highest standards of legal integrity in every action. The American Bar Association's rules govern lawyers, and this vendor's own agreement says it is not acting as one, so the claim asserts compliance with a framework the terms say does not apply to it. No jurisdiction limits are stated. Verified 2 September 2026.
AI Governance and Bias Disclosure
Published governance over model behaviour: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
One transparency commitment is published and no machinery stands behind it. The anti-counterfeit page states that a customer will always have full visibility into when and how AI is used to enforce and protect their brand, which is a governance principle rather than a feature and is more than most records in this corpus offer. The impersonation material adds that every classification carries its reasoning and a risk score, which is explainability at the level of an individual decision. Beyond that nothing was located. No responsible AI page, framework or set of principles exists, no individual or function is named as accountable for model behaviour, nothing describes what is evaluated before a model change ships, and there is no certification such as ISO 42001. Nothing anywhere addresses uneven output, which is a live question on a system operating across 118 countries and 1,500 channels where detection quality plausibly varies by language, script, marketplace and product category, and where a false positive is not an inconvenience but a takedown against a real business. Verified 2 September 2026.
AI Safety and Data Stewardship
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
A trust centre exists and almost nothing beneath it is stated in a way a reviewer could test. Section 1.3 of the terms commits Marq Vision to using information security techniques consistent with good industry standards, and the privacy policy states that all appropriate technical, organisational and administrative security measures are used, both of which are reasonableness language rather than a described control set. Retention is acknowledged and never quantified, with the privacy policy saying data is kept no longer than reasonably necessary and the terms containing no deletion obligation for customer property on termination. No subprocessor is named anywhere: the privacy policy describes agents and affiliated businesses generically. No incident or breach notification practice was located on any surface. Access control is addressed only as the customer's responsibility to protect its own password. A Vanta-hosted trust centre is linked from the footer and would ordinarily carry the control detail; it renders client-side and returned no readable content, which is recorded as a retrieval limit rather than an absence. Verified 2 September 2026.
AI Liability and Recourse
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
A complete allocation is published and the exposure the product actually creates sits entirely with the customer. Section 10.1 caps Marq Vision's aggregate liability at amounts paid in the twelve months before the occurrence, and section 10.2 excludes consequential damages mutually with indemnification obligations carved out. Section 9.1 gives the customer a real indemnity for third-party claims that the platform infringes intellectual property, and section 8.2 warrants performance in compliance with law, non-infringement of the platform, and absence of malicious code. That is more structure than most records here. The limitation worth a buyer's full attention is what the indemnity excludes. Section 9.1(d) removes from Marq Vision's indemnity any claim arising from the customer's use of the services to request removal of content or to send soft notices, which is the core function of the product, and section 9.2 requires the customer to indemnify Marq Vision for actions taken at its instruction and for any consequence of a party being wrongly omitted from the whitelist. Section 8.4(b) states expressly that Marq Vision is in no event responsible for enforcement against an entity that should not have been targeted if it was not properly whitelisted. Wrongful takedown liability, the risk the vendor's own published material identifies as the central one, is allocated to the brand. Verified 2 September 2026.
Practice Systems Integration Depth
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
No integration into the systems legal or IP work already lives in was located, and none is claimed. Checked the home page, the anti-counterfeit, brand protection, content protection, impersonation and MarqLaw pages, the MARQ AI terms in full, the privacy policy in full and the complete site footer on 2 September 2026. No IP management or docketing system is named, no case management or document management system, no connector list, no API or developer documentation, and no integrations page exists in the navigation or the footer. What the platform does connect to is the enforcement estate rather than the customer's own systems, with the privacy policy noting that third-party account credentials may be provided so that account information transmits into the customer's MarqVision account, which is marketplace access rather than workflow integration. Trademark management is delivered through a separate product on a separate domain, and nothing describes data moving between it and the enforcement platform. Verified 2 September 2026.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Nothing published addresses where the platform runs or how customers are separated within it. Checked the home page, the anti-counterfeit, brand protection, content protection, impersonation and MarqLaw pages, the MARQ AI terms in full and the privacy policy in full on 2 September 2026. No cloud provider is named, no hosting region or country is stated, no residency option is offered, nothing distinguishes processing from storage, and no tenancy or isolation model is described. The only geography published is corporate, being offices in Los Angeles, San Francisco, New York, Seoul, Paris, Shanghai and Tokyo, and the only cross-border statement is the privacy policy's reference to the EU-U.S. Privacy Shield Framework. That reference is itself a problem for this axis rather than an answer to it, since Privacy Shield was invalidated in 2020 and replaced in 2023, and the policy carrying it took effect in February 2026. For a platform operated from Delaware with offices in Seoul, Shanghai and Paris, processing brand and seller data across 118 countries, the absence of any residency statement is a material gap. Verified 2 September 2026.
Security Certifications and Trust Center
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
A trust centre is linked and no attestation is stated on any page a reader can render. The footer carries a direct link to a Vanta-hosted trust centre for MarqVision, which is a real route rather than a badge image, and its existence indicates a compliance programme being monitored. What is absent is everything the reader needs from it. No standard is named anywhere on the site itself: neither SOC 2 nor ISO 27001 appears on the home page, the product pages, the terms or the privacy policy, so a buyer learns of a certification only by leaving the site. No auditor, certificate number, report period, scope or trust services criteria is published, and no penetration testing or vulnerability programme is described. The Vanta portal renders client-side and returned no readable content on 2 September 2026, which is recorded as a retrieval limit on this reading rather than as an absence on the vendor's part, and the grade rests on what the vendor states on its own surfaces, which is nothing. Verified 2 September 2026.
Model Supply Chain Disclosure
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
The capabilities are named in proprietary terms and nothing identifies what sits underneath. The published vocabulary is extensive: Atomic Product Detections described as using generative AI, a multi-agent AI that determines whether infringement is valid, an LLM framework auto-generating keywords in any language, image recognition and text analysis, and models stated to be trained on successful enforcement data. Not one of those is attributed. No model is named, no version, no provider entity, and nothing states whether any component is built in-house or reached through a third party, which is the question a brand handing over its product catalogue and seller data would ask. No subprocessor list exists in the terms, the privacy policy or anywhere else, the privacy policy describing third parties only as agents and affiliated businesses. Nothing states where inference runs, and no commitment exists to notify customers when the model set changes. Trademarked capability names in place of provider identification is what this band describes. Verified 2 September 2026.
Commercial Transparency
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
The structure is published in the terms in real detail and no figure appears anywhere. Fees sit on an order form and are stated at section 3.1 to be based on the services purchased rather than actual usage, and to be non-refundable, which tells a buyer this is a committed subscription rather than a consumption model. The term renews automatically for successive twelve month periods under section 7.4 with thirty days notice of non-renewal. Two escalation provisions are published and one is unusually candid: section 3.5 states plainly that fees may be revised on renewal, resulting in an increase every renewal term, and section 3.6 caps increases on multi-year agreements at seven per cent on each anniversary unless the order form says otherwise. Late payment attracts one per cent per month with suspension available after fifteen days. Coverage scope is itself commercial, with the order form specifying regions and section 8.4(a) reserving discretion over which marketplaces are prioritised. What is absent is any number, tier or unit: no pricing page exists, and the entry routes are a free brand scan and a demo request. Verified 2 September 2026.
Firm and Practice Coverage
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
Coverage is documented by industry and by right rather than by buyer role, and on its own terms it is thorough. Seven industry pages are published for beauty, fashion, automotive, pharmaceuticals, food and beverage, network marketing and natural health products, each a category with distinct counterfeiting patterns. Six solution lines are separately documented. The rights covered are stated explicitly, with the anti-counterfeit page saying the platform protects trademarks, copyrights, patents and designs, and geographic reach is given precisely as more than 1,500 channels across 118 countries with offices in seven cities and the site published in English, Korean and Japanese. What is missing is the buyer. There is no page addressed to legal, IP counsel or brand protection teams as roles, and the named individuals in customer stories are overwhelmingly commerce and sales, with a single general counsel among them, so a reader cannot tell from the site whether the platform is bought by the legal function or by digital commerce. No portfolio size band is stated and nothing describes where coverage stops. Verified 2 September 2026.
Legal Signals
What each signal meansA signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
Public material states that customer content trains, refines or personalises models, with no matching term located in the published agreement. Any de identification, anonymisation or aggregation qualifier is recorded in the summary.
Training is described in product material and is not addressed in the agreement, which is what puts it in this value rather than a contractual one. The content protection page states the quoted position, describing an anti-piracy method built on models trained on successful enforcement data. A published article on false positives goes further, describing a tiered human-in-the-loop model in which borderline cases route to analysts and their decisions are then fed back into training to shrink the review queue over time. Enforcement data and analyst decisions are generated in the course of work performed on customer matters using customer-supplied product data, so the material customers contribute is in the training loop even though neither statement uses the phrase customer data. The agreement does not name training at all. Section 2.5(b) of the MARQ AI terms grants a broad right to use information the customer submits, expressly including information about the customer's customers, distributors, inventory and products, for improving platform performance, developing updates and improvements, and improving Marq Vision's and its subsidiaries', vendors' or other customers' business models and services, and separately allows any anonymised data to be used for developing new products. That is a wide improvement right that does not name training, so it does not by itself move the value.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Retention is acknowledged in public materials with no stated period.
Retention is acknowledged and never quantified. The privacy policy states that personal information is kept as long as necessary to fulfil the purpose for which it was collected or to comply with legal or regulatory requirements, and no longer than reasonably necessary, with no period attached to any category. The terms are silent on the point in both directions: section 7.3 sets out the effect of termination as the licence ending and the customer ceasing use and destroying documentation, and imposes no deletion or return obligation on Marq Vision for customer property. Nothing states how long detection records, evidence packages, seller intelligence or enforcement history are held, and no retention setting is offered. Two provisions push the other way and belong on the record. The privacy policy states that some information may remain in records after account deactivation, and that aggregated data derived from or incorporating personal information may continue to be used even after the customer updates or deletes it. Section 4.1 of the terms confirms Marq Vision owns anonymised data outright.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
No located public material addresses walls or matter level segregation.
Checked the home page, the anti-counterfeit, brand protection, content protection, impersonation and MarqLaw pages, the MARQ AI terms in full and the privacy policy in full on 2 September 2026. Nothing describes segregation between customers or isolation of one brand's data from another's. No tenancy model is stated and no access boundary is described beyond the customer's own obligation to protect its password. The question has a specific edge on this product that the published material does not reach. Section 2.5(b) of the terms permits use of information a customer submits, including data about its customers, distributors, inventory and products, to improve the business models and services of Marq Vision's other customers, and the platform maintains a global seller database of more than ten million entries built from enforcement work. A brand would reasonably want to know what separates its catalogue and channel intelligence from a competitor's on the same platform, and nothing published answers that.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Terms commit to notice where lawfully permitted. No transparency report located.
The commitment is in the operative agreement and it is more specific than most. Section 5 of the MARQ AI terms permits disclosure of confidential information where required by valid court order, judicial process or regulatory authority, and then requires the receiving party to use best efforts to preserve confidentiality and to give the quoted notice promptly, unless restrained from doing so by court order. Naming the nature, scope and contents as what must be disclosed to the customer is more than the bare notification most records here offer. The obligation is mutual and survives termination. Two limits belong on the record. Anonymised data is excluded from the definition of confidential information at section 5(d), so the notice commitment does not reach it. And the privacy policy takes a different and broader position for personal information, reserving the right to access, read, preserve and disclose any information Marq Vision believes necessary to comply with law or court order, to enforce its agreements, or to protect its own rights, property or safety, with no notice commitment attached.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Sources are identified without stating the licence or rights basis.
The sources are identified by platform and scale and no rights basis is stated for any of them. Coverage is published as more than 1,500 marketplaces, social media platforms, websites and app stores across 118 countries, with individual platforms named throughout the product material including Amazon, Alibaba, Taobao, TikTok, BiliBili, YouTube, Google and Meta, and a global seller database of more than ten million entries. The customer's own genuine product data is the comparison set for SKU-level matching. What is absent is the legal basis on which any of it is gathered. Nothing states whether marketplace listings are collected under platform agreements, APIs, brand registry programmes or by crawling against those platforms' terms of use, and the distinction matters commercially as well as legally, since the company separately markets its marketplace partnerships in Southeast Asia and China and holds direct takedown authority on YouTube, which implies negotiated access on some channels and not others. No licence, agreement or permission is described for the underlying listing data anywhere.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
No located public material addresses whether authority is checked for subsequent history.
Checked the home page, the anti-counterfeit, brand protection, content protection, impersonation and MarqLaw pages, the MARQ AI terms and the privacy policy on 2 September 2026. No public material addresses subsequent history, treatment or good law checking, and none is claimed. The product does not retrieve primary law, so a citator has nothing to operate on and the limb does not bite in its usual form. The analogous currency question in this product class is whether the intellectual property right being enforced is still valid and in force, and it is addressed only as the customer's responsibility rather than as a system check: section 8.3(d) of the terms requires the customer to warrant that it is the rightful owner or licensee of all rights subject to the services, and section 8.4(b) states that the sufficiency of the customer's own registrations is outside Marq Vision's control. Nothing describes the platform verifying a registration's status before enforcing on it.
Refusal and Uncertainty Behaviour
What does the product do when the answer is not in the corpus?
The product exposes a confidence or grounding score without an explicit abstention path.
Confidence is scored and surfaced, and nothing describes the system declining. The impersonation material states that every classification produced by the multi-agent system comes with its reasoning and a risk score, so a user sees both a rationale and a graded confidence rather than a bare determination. A published article on false positives adds that a tiered human-in-the-loop model routes only borderline cases to analysts, which is a confidence threshold operating on the workflow even though the threshold itself is never defined. What is absent is any account of what the system does when it cannot reach a determination at all, whether it abstains, holds an item, or defaults to escalation, and nothing states what confidence level is required before an automated enforcement action proceeds. The customer-maintained whitelist is a suppression mechanism supplied by the customer rather than a behaviour of the model.
Fabricated Citation Record
Does a public court record exist involving output from this product?
No court order, opinion or disciplinary record naming this product has been located as of the date shown. This is a statement about the public record, not a finding about the product.
Searched the AI Hallucination Cases database maintained by Damien Charlotin, and reporting drawing on it, on 2 September 2026 on the product and corporate name MarqVision and Marq Vision Inc. No court order, opinion or disciplinary record naming the product was located. This is a statement about the public record rather than a finding about the product. The signal fits this product class poorly and the reason is worth recording: MarqVision generates takedown notices and enforcement packages rather than legal citations, so its characteristic failure is a wrongful enforcement action against a legitimate seller rather than a fabricated case reference in a filing. The vendor's own published material identifies that risk directly, naming misrepresentation liability under 17 U.S.C. 512(f), tortious interference and platform reporting suspension as the consequences of false positives. Those outcomes surface as civil claims or platform sanctions rather than as entries in a hallucination tracker, and no tracker indexes them.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Public materials refer to professional responsibility in general terms without naming guidance.
An authority is named and no instrument is, and the claim sits awkwardly against the agreement. The anti-counterfeit page states that MarqVision's AI-driven brand protection solutions comply fully with the ABA's ethics guidelines, ensuring the highest standards of legal integrity in every action. Naming the American Bar Association puts this above a bare gesture at professional standards, which is why the lowest value does not fit. But no rule, model rule or formal opinion is identified, nothing states which obligations are said to be met or how compliance is assessed, and ABA Formal Opinion 512 on generative AI is not referenced. The claim also runs against section 1.2 of the company's own terms, which states that Marq Vision does not and cannot provide legal guidance or advice and that no attorney-client relationship arises, since the ABA's rules bind lawyers rather than software vendors. A reader is left with an assertion of compliance with a framework the agreement says does not govern the service.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure.
Time and cost savings are published with figures and nothing addresses the bill. The company states that Marq AI saves its expert teams 180 times the time previously required to enforce infringements, that enforcement runs 50 per cent faster than traditional brand protection companies, and that mean time to resolution is fifteen minutes, and it publishes a Saturation Rate metric and revenue recovery framing intended to show enforcement paying for itself. Nothing addresses how AI-assisted work is recorded, billed or disclosed, and no per matter record of AI-assisted work was located. The signal's usual direction is partly inverted here, since the buyer is the brand owner rather than a firm billing a client, so compressed enforcement time accrues to the customer rather than changing an invoice. It is not fully inverted: MarqLaw engages partner law firms who bill for litigation, customs and field investigation work supported by platform output, and nothing published describes how AI-assisted preparation is reflected in those engagements.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
No located public material supports a client side disclosure obligation.
Checked the MARQ AI terms in full, the privacy policy in full, the home page and the anti-counterfeit, brand protection, content protection, impersonation and MarqLaw pages on 2 September 2026. No subprocessor list exists in any form. The privacy policy describes third parties only by category, covering agents performing tasks on Marq Vision's behalf, affiliated businesses and third-party law firms, advertisers and partners receiving de-identified data, without naming a single entity. No model or AI provider is named anywhere on any surface. No data processing agreement is published, and no forwardable client-facing disclosure material was located. A Vanta-hosted trust centre is linked from the footer and would be the natural home for a subprocessor list; it renders client-side and returned no readable content, which is recorded as a retrieval limit rather than an absence. A brand asked by a partner or regulator which entities process its catalogue and seller data could not answer from anything published.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
This is the closest any record in this pull comes to the signal without meeting it. Evidence capture is a published product function rather than a by-product: the anti-counterfeit page states that the platform automatically extracts and stores the evidence needed to shut down illicit activity and generates full self documentation reports, MarqLaw is marketed on courtroom-ready evidence packages, the impersonation material describes a report-ready enforcement package prepared for each channel, and a published article states that consistent enforcement backed by a documented record of every action is what redirects counterfeiters elsewhere. The anti-counterfeit page also carries an AI transparency commitment, stating that a customer will always have full visibility into when and how AI is used to enforce and protect their brand. What is not described is the record this signal asks for: nothing states that the visibility is exportable, and nothing captures which model or agent made a given determination, on what confidence, or which human reviewer confirmed it, which is precisely what a party challenging a takedown or a 512(f) claim would seek.