Summary: The brand protection industry is moving from notice-processing to intelligence work, driven by tightening platform access for external scrapers, expanding platform-internal scraping aimed at competitors, and AI capabilities powerful enough to make repeat-infringer enforcement defensible at scale.

The paradox is that the same AI clarity that lets rights owners build cases against repeat infringers will reveal a scale of tolerated infringement so commercially significant that platforms will be structurally reluctant to act on it.

The actors on both sides who recognise this early — rights owners by combining evidence with regulatory leverage, and platforms by choosing genuine seller-level enforcement over compliance optics — will define what the industry looks like on the other side of the decade.

The Future of Brand Protection

A Hypothesis on the Coming Decade of IP Enforcement, Platform Behaviour, and Artificial Intelligence

by Claude, Research mode. Warning, LLMs may hallucinate!

Introduction

The brand protection industry stands at an inflection point. For the better part of two decades, the playbook has been familiar: rights owners and their vendors crawl marketplaces and social commerce sites, identify infringing listings, file notices, and watch most of those listings reappear within days under different seller identities. The model has been stable, profitable for vendors, and quietly insufficient for rights owners. That stability is about to break.

Three forces are converging that will reshape the industry over the next five to ten years: a tightening of platform access for external crawlers, a parallel expansion of platform-internal scraping aimed at competitors, and the arrival of AI capabilities powerful enough to make repeat-infringer enforcement scalable for the first time in the industry’s history. Each of these forces, taken alone, would meaningfully alter the field. Together, they create a paradox that will define how brand protection evolves, who profits from it, and whether rights owners actually end up better protected at the end of the decade than they are today.

This hypothesis lays out where those forces are likely to lead, where they are likely to collide, and what rights owners and platforms each need to understand if they want to avoid the worst outcomes.

Force One: The Closing of the Open Marketplace

E-commerce platforms have always had a complicated relationship with external scrapers. Brand protection vendors, market intelligence firms, price aggregators, and academic researchers have for years relied on the relative openness of marketplace listings to do their work. That openness is eroding, and the erosion is set to accelerate.

There are several reasons for this, and only some of them are stated publicly. The publicly cited reasons tend to focus on infrastructure load, user privacy, and the protection of seller data. The less-stated reasons are more important. External scrapers, when used by brand protection vendors and investigative journalists, produce something platforms increasingly do not want produced: independent, defensible quantification of how much infringing activity is actually taking place. If a third party can crawl a marketplace and demonstrate that a particular seller has been listing counterfeit goods for three years across forty product variants and hundreds of thousands of dollars in estimated sales, that finding is harder to dismiss than an internal report a platform can shape, filter, or simply not produce.

Expect platforms over the coming years to introduce progressively more aggressive technical measures: stricter rate limits, mandatory authentication, behavioural fingerprinting, legal action against scrapers under computer-misuse statutes, and the funnelling of brand protection vendors into official APIs whose data scope is controlled by the platform itself. The official APIs will be marketed as a partnership offering. In practice they will be a way of deciding which questions can be asked and which cannot.

The strategic consequence for brand protection vendors is profound. The industry has been built on the assumption that the marketplace is a public space that can be observed. As that assumption weakens, vendors will need to invest in alternative evidentiary techniques, including transactional test purchases at scale, supply-chain mapping, customs and import data, social-graph analysis of seller networks, and direct partnerships with payment processors and logistics providers. The vendors that survive the next decade will be the ones that can build a picture of an infringing seller without ever needing to scrape a single product page.

Force Two: The Asymmetry of Scraping

At the same time as platforms close their gates to external scrapers, they will dramatically expand their own internal scraping operations directed at competitors. This is already happening, and the trend will intensify.

Every major marketplace now operates a sophisticated competitive intelligence function whose job is to scrape rival platforms for pricing, assortment, advertising creative, search-result composition, seller migration patterns, promotional cadence, and review velocity. The data feeds pricing engines, ad-auction strategy, category management, seller acquisition campaigns, and the algorithms that decide which products surface where. The asymmetry is the point: a platform wants total visibility into its competitors and total opacity around itself.

This asymmetry has two consequences for brand protection. First, it sets a precedent that platforms cannot credibly disown when challenged on their own access policies. A platform that scrapes its competitors at industrial scale while suing a brand protection vendor for crawling its own listings is making an argument that will not survive close legal or regulatory scrutiny indefinitely. Expect rights-owner trade associations and competition regulators to begin pressing this contradiction. Second, the same scraping infrastructure platforms build for competitive intelligence is structurally identical to the infrastructure they would need to identify infringing sellers on their own surfaces. The technical capacity exists. Whether platforms choose to point it inward, and what they do with what they find, is a question of will rather than capability.

Force Three: The AI Paradox

Artificial intelligence will transform brand protection more decisively than any technology since the notice-and-takedown regime itself. The transformation will not be uniform across the value chain, and that unevenness is the source of the central paradox of the next decade.

The Promise: Clarity at Scale

For rights owners and brand protection vendors, AI offers something the industry has never had: the ability to look across millions of listings, hundreds of thousands of seller accounts, and years of historical data and identify the patterns that distinguish a one-time mistake from a persistent bad actor. Multi-modal models can recognise infringing imagery even when sellers alter fonts, swap colours, change packaging shots, or use AI-generated product photos to evade narrow trademark-matching algorithms. Language models can read product descriptions, customer questions, and review threads in dozens of languages and surface the linguistic and metadata signals that reveal a seller’s true identity, network affiliations, and infringement history.

More importantly, AI can do something that human reviewers and rule-based systems have struggled to do: it can reason across siloed data. A model with access to listing data, seller account data, transaction patterns, shipping origins, payment instruments, and historical notice records can construct a coherent picture of a repeat infringer that no individual reviewer, working within a narrow Standard Operating Procedure, would ever assemble. The brand protection industry will spend the next several years building exactly these capabilities, and the rights owners who invest in them first will gain a decisive evidentiary advantage.

This is the optimistic half of the paradox. For the first time, rights owners will be able to walk into a platform escalation meeting with a defensible, quantified case that a specific seller has generated a specific volume of infringing sales over a specific period, supported by evidence that does not depend on the platform’s own filtered reporting. The era of vague complaints and disputed numbers will end.

The Problem: Enforcement at Scale Has a Price Tag

Here is where the paradox sharpens. The same AI capability that lets rights owners identify repeat infringers at scale will also reveal, with uncomfortable precision, just how much of a platform’s gross merchandise volume is attributable to those infringers. Industry-wide, the number is almost certainly larger than platforms publicly acknowledge and larger than many rights owners themselves currently estimate.

Once that number becomes visible and defensible, platforms face a choice that is both operational and financial. Enforcing seriously against repeat infringers at the scale AI makes possible would mean removing thousands of high-volume sellers, deindexing their inventory, refunding affected buyers in some jurisdictions, and absorbing the loss of advertising revenue, fee revenue, and category breadth that those sellers represent. The bottom-line impact is not theoretical. For some categories on some platforms, the share of revenue connected to sellers with substantial infringement histories is high enough that decisive enforcement would show up directly in quarterly results.

This is the structural reluctance that will define platform behaviour in the AI era. Platforms will not announce that they are slowing enforcement. They will instead do what the incentive structure quietly encourages: invest in tools that look impressive in public reporting while preserving the underlying flow of seller volume. Expect to see more dashboards, more transparency reports, more brand-portal features, and more press releases about enforcement partnerships, alongside a quiet persistence of the same repeat infringers who have been operating for years.

The Hidden Architecture of Platform Reluctance

The paradox above is not a hypothetical. It is the predictable extension of patterns that already shape how large e-commerce and social-commerce platforms handle enforcement today, and any honest hypothesis about the future has to take those patterns seriously.

Inside major platforms, IP enforcement is often structurally biased toward processing individual notices rather than removing the sellers behind them. A seller can accumulate thousands of validated notices and remain active because each notice is treated as a discrete compliance event rather than as evidence of a persistent bad actor. Listings get removed; the seller account does not. Internal seller-education programmes can function, in effect, as a way of resetting the enforcement clock without addressing the underlying behaviour. Notice histories may decay or age out of the system entirely, making cumulative cases against repeat offenders progressively harder to build the longer the offender survives.

Detection systems often rely heavily on rights-owner-supplied trademarks and word marks, while broader signals, such as seller behaviour, listing metadata, image manipulation, packaging cues, and cross-account network patterns, sit unmonitored in backend tables that enforcement teams may not even have dashboards for. Notice data, seller data, listing data, and transaction data frequently live in separate systems. The internal teams who would need to connect those datasets to identify a repeat infringer often cannot easily do so, and the teams who could, such as anti-fraud, anti-piracy, brand protection, legal, and law-enforcement liaison, are often kept in separate organisational silos with limited cross-coordination.

The friction does not stop at architecture. Reviewers are frequently outsourced, undertrained on IP and cybercrime patterns, and measured on throughput rather than accuracy or pattern detection. Strict Standard Operating Procedures can leave a seller technically within policy even when the cumulative evidence of abuse is overwhelming. Employees inside platforms who try to do enforcement properly often face delays of months getting access to relevant data, must build their own dashboards from scratch, and discover that internal reporting is filtered or selectively presented in ways that understate the true scale of the problem. Raising those findings can carry real professional cost. Specialists who joined the work because they wanted to stop bad actors find themselves processing notices instead, and many leave.

The marketplace and social-commerce sides of integrated platforms are often siloed by design, creating enforcement blind spots that sophisticated sellers learn to exploit. Counterfeit and design-right infringement that does not map cleanly to a trademark or word mark can fall through the gaps entirely. Crude physical counterfeits, photocopied books, altered fonts, three-ring-bound knock-offs, and digital download variants of the same infringing content can persist for years against the same seller, particularly where independent creators and smaller publishers, who are commercially attractive but lack enforcement leverage, are the targets.

None of this is conspiracy. Most of it is the predictable outcome of safe-harbour incentives that may discourage proactive monitoring, fragmented organisational structures, narrow legal-compliance framings of what enforcement even means, and the genuine difficulty of running enforcement at the scale of modern marketplaces. But it is the substrate on which the AI-driven future of brand protection will be built, and ignoring it would be a serious mistake.

What This Means for Rights Owners

Rights owners who plan their brand protection strategy for the next decade on the assumption that platforms will respond rationally to better evidence are likely to be disappointed. The evidence will get better. The response will not automatically follow.

Effective IP enforcement in the AI era will require rights owners to do several things they have historically been reluctant to do.

What This Means for Platforms

Platforms reading their own incentives narrowly will conclude that the AI era is something to be managed defensively: tighten access, control the API, publish good reports, keep the seller volume flowing. That conclusion is short-sighted, and the platforms that act on it will pay a price that will become visible over the second half of the decade.

The same AI capabilities that let rights owners build defensible cases against repeat infringers will also be available to regulators, journalists, plaintiffs’ lawyers, and class-action firms. The era in which a platform could rely on the difficulty of independent quantification to manage its public narrative is closing. When the scale of tolerated infringement on a given platform becomes possible to demonstrate from the outside, the regulatory, reputational, and litigation consequences will arrive together. Platforms that have invested in serious seller-level enforcement, transparent reporting that reflects internal reality rather than filtered narratives, and genuine cross-team coordination between brand protection, anti-fraud, and law-enforcement liaison functions will weather that period. Platforms that have invested in optics will not.

The specific risks platforms should be planning against include a meaningful tightening of safe-harbour interpretations in major jurisdictions, the increasing willingness of courts to look behind platform self-reporting at internal data, the growing capability of rights-owner coalitions to bring coordinated multi-jurisdictional action, and the reputational risk that arrives when a major investigative outlet publishes an AI-driven analysis showing that named sellers operated for years against thousands of notices. Each of these risks is foreseeable. None of them is adequately addressed by better dashboards alone.

The Shape of the Industry in 2030

The brand protection industry of the next decade will look meaningfully different from the one rights owners and vendors are operating in today.

Vendors will bifurcate. A first tier will compete on AI capability, evidentiary depth, and the ability to operate without depending on platform-supplied data feeds. They will look more like investigative intelligence firms than like the takedown-volume vendors of the 2010s. A second tier will continue to provide notice-processing services, increasingly commoditised and increasingly mediated through platform APIs, and will compete largely on price. The middle of the market, vendors who lack distinctive intelligence capability but charge premium prices, will compress.

Platforms will continue to invest in brand protection optics while quietly preserving the seller volume that drives revenue, until external pressure makes that posture untenable. The first major platform to genuinely commit to seller-level enforcement, repeat-infringer termination thresholds that bite, integration of seller and notice data, and transparent reporting that reflects internal reality will gain a meaningful regulatory and reputational moat. Whether any platform chooses that path before being forced to is one of the open questions of the decade.

Rights owners will increasingly run their brand protection function as an intelligence operation rather than a compliance one. The largest brand owners will build internal teams that combine IP, cybercrime, supply-chain analysis, and data science capability, and will treat their external vendors as evidence-generation partners rather than as outsourced takedown operators. Smaller rights owners and independent creators, who are the most exposed to repeat infringement and the least equipped to defend against it, will benefit most from collective action through trade associations and from the increasing availability of AI tooling that lowers the cost of evidentiary work.

Regulators, courts, and law-enforcement bodies will become more central to the picture than they have been since the early years of e-commerce. The combination of AI-driven evidence, regulatory appetite for action against tolerated infringement, and the increasing political salience of counterfeit and pirated goods will produce the conditions for a meaningful shift in how platform liability is interpreted in major jurisdictions. The shift will not be uniform, and platforms will fight it, but the direction is set.

Conclusion: The Warning in Both Directions

The hypothesis above can be summarised as a warning that runs in two directions at once.

To rights owners: the assumption that better evidence automatically produces better enforcement is wrong, and acting on it will leave you exposed. AI will give you clarity you have never had before. Platforms will, by default, not respond to that clarity by cleaning up their marketplaces, because the financial cost of doing so is real and the incentive to absorb that cost is weak. Effective enforcement in the next decade will require you to combine evidentiary capability with regulatory leverage, coordinated industry action, law-enforcement engagement, and a willingness to litigate. Rights owners who treat brand protection as a compliance function rather than as an intelligence-and-pressure function will spend the decade watching their evidence accumulate while their losses do too.

To platforms: the assumption that the difficulty of independent quantification will continue to give you narrative control is also wrong. The infrastructure is being built, by your competitors’ adversaries and increasingly by regulators themselves, to demonstrate from outside the walls what your internal teams have been telling you for years. Tolerating repeat infringers because each individual termination is operationally inconvenient and commercially costly is a defensible position only as long as the cost of toleration is hidden. That cost is about to become visible. Platforms that confuse compliance optics with enforcement, that keep their anti-fraud, brand protection, anti-piracy, legal, and law-enforcement liaison teams in separate silos, that filter their internal reporting to support a preferred narrative, and that resource their enforcement function for throughput rather than for genuine pattern-detection and seller-level action, will find that the regulatory and reputational reckoning, when it arrives, is harder to manage than serious enforcement would have been.

Brand protection is moving from a notice-processing industry to an intelligence industry. The transition will be uneven, contested, and at times openly adversarial between rights owners and the platforms they depend on. But the direction is clear. The actors on each side who recognise it early, and act on the recognition rather than on the comfort of the existing model, will define what the industry looks like on the other side of the decade.

Public Sources Supporting Three Central Forces in the Future of Brand Protection

TL;DR

• Force 1 (platforms tighten gates): — landmark scraping litigation (hiQ v. LinkedIn, Meta v. Bright Data, X Corp v. Bright Data), an industry-wide arms race in anti-bot defences (with automated bot traffic surpassing human-generated traffic for the first time in a decade at 51% of all web traffic in 2024 per Imperva’s 2025 Bad Bot Report), and Amazon’s evolution toward a controlled Brand Registry/Project Zero/Transparency portal model all point in the same direction. https://cpl.thalesgroup.com/about-us/newsroom/2025-imperva-bad-bot-report-ai-internet-traffic , https://www.imperva.com/resources/resource-library/reports/2025-bad-bot-report/

• Force 2 (platforms scrape rivals): — the FTC’s Amazon antitrust complaint (Project Nessie), the WSJ “Big River”/Project Curiosity exposé, the European Commission’s Amazon seller-data case, and academic work on the asymmetry of CFAA-driven anti-scraping enforcement directly evidence the asymmetry; named-program journalism is thinner outside Amazon.

• Force 3 (AI paradox): — vendor and academic work demonstrate AI-enabled multi-modal counterfeit detection and seller-network mapping, while OECD/EUIPO, USTR, GAO, EU DSA proceedings against Temu and Shein, and the long Tiffany v. eBay safe-harbour line illustrate platforms’ structural reluctance to act decisively against repeat infringers.

Key Findings (by Force)

FORCE ONE — E-commerce platforms are tightening access for external bots, scrapers, and brand protection vendors

1. hiQ Labs, Inc. v. LinkedIn Corp. — Ninth Circuit Opinion (April 18, 2022) and Stipulated Permanent Injunction (Dec. 6, 2022) | Publisher: U.S. Court of Appeals for the Ninth Circuit / N.D. Cal.; client-alert summary by Jenner & Block | URL: https://www.jenner.com/en/news-insights/publications/client-alert-data-scraping-in-hiq-v-linkedin-the-ninth-circuit-reaffirms-narrow-interpretation-of-cfaa . The Ninth Circuit held that scraping public data is not “without authorization” under the CFAA, but on remand the district court enforced LinkedIn’s contractual no-scraping terms and entered a $500,000 stipulated judgment plus permanent injunction — illustrating the shift from CFAA to contract-based enforcement to bar scrapers. https://www.privacyworld.blog/2022/12/linkedins-data-scraping-battle-with-hiq-labs-ends-with-proposed-judgment/

2. “LinkedIn’s Data Scraping Battle with hiQ Labs Ends with Proposed Judgment” | Publisher: Privacy World (Squire Patton Boggs), Dec. 2022 | URL: https://www.privacyworld.blog/2022/12/linkedins-data-scraping-battle-with-hiq-labs-ends-with-proposed-judgment/ . Documents the November 2022 N.D. Cal. summary judgment finding LinkedIn’s user-agreement scraping prohibition enforceable and the December 2022 consent judgment that ended hiQ — confirming platforms can shut down scrapers contractually even after the Ninth Circuit’s narrow CFAA reading.

3. Meta Platforms, Inc. v. Bright Data Ltd., 3:23-cv-00077-EMC (N.D. Cal., Jan. 23, 2024) — Quinn Emanuel client alert | Publisher: Quinn Emanuel; date: Jan. 24, 2024 | URL: https://www.quinnemanuel.com/the-firm/news-events/client-alert-meta-v-bright-data-significant-decision-for-web-scraping-industry/ . Judge Edward Chen granted Bright Data summary judgment, holding Meta’s terms barred logged-in but not logged-off scraping of public data; Meta dropped the remaining claims weeks later (TechCrunch, Feb. 26, 2024, https://techcrunch.com/2024/02/26/meta-drops-lawsuit-against-web-scraping-firm-bright-data-that-sold-millions-of-instagram-records/ ). Shows platforms aggressively pursuing scrapers even where they ultimately lose. https://www.fbm.com/publications/major-decision-affects-law-of-scraping-and-online-data-collection-meta-platforms-v-bright-data/ , https://brightdata.com/blog/web-data/court-rules-in-favor-of-bright-data-in-meta-v-bright-data-case , https://research.aimultiple.com/is-web-scraping-legal/ , https://brightdata.com/blog/general/meta-dismisses-claim-against-bright-data , https://nubela.co/blog/meta-lost-the-scraping-legal-battle-to-bright-data/

4. “Proskauer Secures Dismissal of Scraping Claims Against Bright Data” (X Corp v. Bright Data, N.D. Cal., May 10, 2024) | Publisher: Proskauer Rose; date: May 23, 2024 | URL: https://www.proskauer.com/release/proskauer-secures-dismissal-of-scraping-claims-against-bright-data . Judge William Alsup dismissed X’s contract, tort, misappropriation, and copyright-preempted claims, observing that letting X enforce its terms against scrapers would “yank into [X’s] private domain… information open to all” and risk “information monopolies that would disserve the public interest” — direct evidence platforms are litigating aggressively to enforce closed gates. https://www.courthousenews.com/judge-tosses-xs-contract-claims-against-data-scraping-company/ , https://www.courthousenews.com/judge-tosses-xs-contract-claims-against-data-scraping-company/

5. “District Court Adopts Broad View of Copyright Preemption in Data Scraping Case” | Publisher: Skadden, Arps; date: May 2024 | URL: https://www.skadden.com/insights/publications/2024/05/district-court-adopts-broad-view . Analyzes the X Corp v. Bright Data ruling and notes that despite courts limiting platforms’ contract claims, X had previously responded to scrapers by limiting tweet visibility for non-logged-in users — concrete platform technical countermeasure.

6. “X Gets Aggressive About Scraping on its Platform” | Publisher: ZwillGen; date: 2023 | URL: https://www.zwillgen.com/data-security/x-gets-aggressive-about-scraping-on-its-platform/ . Catalogs three near-simultaneous X Corp lawsuits — against John Doe IP addresses, Bright Data, and the Center for Countering Digital Hate — and details X’s allegations that scraping causes “heavy load” on servers and that X uses “anti-scraping technology” actively evaded by scrapers. Shows platforms attacking even public-interest researchers and watchdog scraping. https://natlawreview.com/article/x-corp-lawsuits-target-data-scraping

7. “2025 Bad Bot Report” | Publisher: Imperva (Thales); date: April 15, 2025 | URL: https://www.imperva.com/resources/resource-library/reports/2025-bad-bot-report/ (and 2024 report: https://www.imperva.com/resources/resource-library/reports/2024-bad-bot-report/ ). Imperva states: “Automated bot traffic surpassed human-generated traffic for the first time in a decade, constituting 51% of all web traffic in 2024,” with bad bots reaching 37% of all internet traffic in 2024 (up from 32% in 2023). The 2024 report further shows that in 2023, “evasive” advanced + moderate bots accounted for 60.5% of bad bot traffic, while in 2024 simple bot traffic grew to 45%. Documents the technical countermeasures (TLS fingerprinting, behavioural analysis, residential-proxy detection) deployed at scale and the AI-driven escalation of the arms race. https://securityboulevard.com/2025/04/2025-imperva-bad-bot-report-how-ai-is-supercharging-the-bot-threat/ , https://cpl.thalesgroup.com/blog/identity-data-protection/imperva-2024-bad-bot-report-insights-and-solutions

8. “How to Bypass Anti-Bot Protection When Web Scraping” | Publisher: Scrapfly; date: 2024 | URL: https://scrapfly.io/blog/posts/how-to-bypass-anti-bot-protection-when-web-scraping . Practitioner-side description of layered platform defences (Cloudflare, DataDome, PerimeterX, Akamai, Kasada) using TLS fingerprinting, behavioural analysis, JavaScript challenges, and proof-of-work — useful for documenting the technical sophistication of platform anti-scraping stacks. https://use-apify.com/blog/web-scraping-anti-detection-2026

9. Aneja & Aneja, “When Handshakes Tell the Truth: Detecting Web Bad Bots via TLS Fingerprints” | Publisher: arXiv preprint; date: 2025 | URL: https://arxiv.org/html/2602.09606v1 . Academic confirmation that protocol-layer TLS fingerprinting (e.g., JA3) is a primary modern bot-detection signal — i.e., platform countermeasures now operate well below the application layer where brand-protection scrapers must compete. https://arxiv.org/html/2602.09606v1

10. “Amazon Strengthens Brand Protection for Trademark Owners: Key Updates in 2025” | Publisher: Dickinson Wright (analyzing Amazon’s Brand Registry roles); date: 2025 | URL: https://www.dickinson-wright.com/news-alerts/blog-arndt-amazon-2025-trademarks ; primary Amazon documentation: https://sell.amazon.com/blog/brand-registry-roles . Describes how Amazon has migrated brand-protection workflows into a “Registered Agent” role gated by Amazon’s own approval, with new 2025 features (Brand Catalog Lock, Project Zero, Transparency QR codes) that channel infringement reporting through Amazon-controlled portals rather than independent scraping. https://sell.amazon.com/blog/brand-registry-roles

11. “How Amazon is Changing the US Trademark System” (interview with Sprigman & Tushnet) | Publisher: New York University; date: July 2025 | URL: https://www.nyu.edu/about/news-publications/news/2025/july/how-amazon-is-changing-the-us-trademark-system.html . Academic interview characterizing Amazon’s Brand Registry as a “shadow trademark system” — gating brand protection access through Amazon’s own rules and de facto creating private rights inside Amazon that displace independent monitoring/quantification.

12. Daniel J. Solove & Woodrow Hartzog, “The Great Scrape: The Clash Between Scraping and Privacy” | Publisher: California Law Review, Vol. 113; date: 2025 | URL: https://www.californialawreview.org/print/great-scrape . Legal-academic analysis explicitly noting that companies use the CFAA “as a means of eliminating competitors whose business models rely on data scraping” and often invoke “privacy” as pretext — directly supports the thesis that platform anti-scraping enforcement is, in part, about controlling who can quantify activity on their surfaces.

FORCE TWO — Platforms scrape competitors while tightening their own gates (the asymmetry)

1. Dana Mattioli & Sara Nassauer, “Inside Amazon’s Secret Operation to Gather Intel on Rivals” | Publisher: Wall Street Journal; date: April 18, 2024 | URL: https://www.wsj.com/business/inside-amazons-secret-operation-to-gather-intel-on-rivals-abb82907 . The single most direct primary source: Amazon’s “Project Curiosity” used a shell company, “Big River Services International,” and Amazon staff posing as third-party sellers on Walmart, eBay, Shopify, FedEx, and other rivals to harvest pricing, logistics, advertising, and catalogue data, with code names (”OnTime Inc.” for FedEx) and paper-only briefings, all regularly briefed up to current Amazon Stores CEO Doug Herrington. Direct evidence of asymmetry.

2. FTC and 17 States v. Amazon.com, Inc., No. 2:23-cv-01495-JHC (W.D. Wash.) — Revised Redacted Complaint | Publisher: U.S. Federal Trade Commission; dates: filed Sept. 26, 2023; partial unredactions Nov. 2, 2023 | URL: https://www.ftc.gov/system/files/ftc_gov/pdf/1910134amazonecommercecomplaintrevisedredactions.pdf . Primary regulatory filing: details Amazon’s “Project Nessie” pricing algorithm, which monitored Walmart, Target, and other rivals’ pricing reactions to Amazon’s price hikes and generated >$1 billion in excess profit; quotes Amazon executives describing a “game theory approach” to “rapidly copying others’ moves to the penny.” Establishes that Amazon’s pricing apparatus depends on continuous competitor surveillance.

3. Dana Mattioli, “Amazon Used Secret ‘Project Nessie’ Algorithm to Raise Prices” | Publisher: Wall Street Journal; date: October 3, 2023 | URL: https://www.wsj.com/business/amazon-used-secret-project-nessie-algorithm-to-raise-prices-6c593706 . Original WSJ reporting that broke the Nessie story; corroborates and contextualizes the FTC’s redacted allegations.

4. “Unredacted FTC suit shows ‘Project Nessie’ price-raising algorithm made Amazon $1.4B” | Publisher: TechCrunch; date: Nov. 2, 2023 | URL: https://techcrunch.com/2023/11/02/unredacted-ftc-suit-shows-project-nessie-price-raising-algorithm-made-amazon-1-4b/ . Useful secondary technical breakdown of how Nessie used predictive monitoring of rival retailers’ algorithmic responses.

5. European Commission, “Antitrust: Commission accepts commitments by Amazon barring it from using marketplace seller data, and ensuring equal access to Buy Box and Prime” | Publisher: European Commission, Press Release IP/22/7777; date: Dec. 20, 2022 | URL: https://ec.europa.eu/commission/presscorner/detail/en/ip_22_7777 (corresponding 2020 Statement of Objections IP/20/2077: https://ec.europa.eu/commission/presscorner/detail/en/ip_20_2077 ). Primary EU regulatory finding that Amazon “systematically” used non-public third-party seller data — fed into Amazon’s automated retail pricing tool — to compete against those sellers; directly evidences regulator-confirmed competitor data harvesting at the platform level. https://www.cuatrecasas.com/en/global/competition-eu-law/art/european-commission-accepts-amazons-commitments-and-closes-its-investigation

6. Khadeeja Safdar, “Amazon Scooped Up Data From Its Own Sellers to Launch Competing Products” (WSJ investigation) | Publisher: Wall Street Journal; date: April 23, 2020 | summary URL: https://www.cnbc.com/2020/04/23/wsj-amazon-uses-data-from-third-party-sellers-to-develop-its-own-products.html ; Retail Dive on SEC follow-up: https://www.retaildive.com/news/amazon-under-investigation-by-sec-over-disclosures-on-seller-data-use-wsj/621735/ . Original WSJ investigation that triggered the Congressional referral and SEC inquiry into whether Amazon executives lied to Congress about using individual seller data.

7. People of the State of California v. Amazon.com, Inc. — unsealed evidence reported in Leah Nylen, “Amazon’s Alleged Price-Fixing Targets Walmart, Home Depot, Chewy” | Publisher: Bloomberg; date: April 20, 2026 | URL: https://www.bloomberg.com/news/articles/2026-04-20/amazon-s-alleged-price-fixing-targets-walmart-home-depot-chewy . Documents internal Amazon training instructing employees not to use email when discussing price-match work; alleges Amazon pressured vendors (Levi’s, Hanes, Newell, Linon, Armen Living) to raise prices on Walmart, Target, Home Depot, and Chewy after Amazon-curated link lists of “styles of concern.” Newest primary-source evidence of platform-on-platform price surveillance.

8. Aaron R. Gott, “Liability for Data Scraping Prohibitions under the Refusal to Deal Doctrine” | Publisher: University of Chicago Law Review, Vol. 87; date: 2020 | URL: https://lawreview.uchicago.edu/print-archive/liability-data-scraping-prohibitions-under-refusal-deal-doctrine-incremental-step . Academic article articulating the asymmetry as a Section 2 antitrust problem: dominant platforms ban scraping by competitors while running analytics services on the same data. Quotes the Ninth Circuit warning that selective bans on scraping by potential competitors “may well be considered unfair competition.”

9. “The 8 Best Walmart Scrapers in 2026” | Publisher: Bright Data; date: 2026 | URL: https://brightdata.com/blog/web-data/best-walmart-scrapers . Industry-side material documenting that scraping Amazon and Walmart pricing/catalogue data is now standard competitive-intelligence practice; useful context — while platforms litigate against scrapers, retailers and brands (and platforms via shell entities) routinely participate in the same scraping market.

10. Solove & Hartzog, “The Great Scrape” | Publisher: California Law Review, Vol. 113; date: 2025 | URL: https://www.californialawreview.org/print/great-scrape . Same source as Force 1 #12; in this context cited for its explicit framing of the hiQ/Bright Data anti-scraping doctrine as a tool platforms use to entrench their dominance while collecting/scraping competitively themselves.

FORCE THREE — The AI paradox in brand protection

A. AI gives rights owners unprecedented clarity

1. Cheung, She, Sun & Zhou, “Detecting Online Counterfeit-Goods Sellers Using Connection Discovery” | Publisher: ACM Transactions on Multimedia Computing, Communications, and Applications; date: 2019 | URL: https://dl.acm.org/doi/10.1145/3311785 . Foundational academic work using deep learning on 473K shared images from Taobao, Instagram, and Carousell to map counterfeit-seller networks; the proposed framework was 30% more accurate than object-recognition baselines, demonstrating multi-modal seller-network mapping.

2. Cheung, She & Liu, “Deep Learning-Based Online Counterfeit-Seller Detection” | Publisher: IEEE INFOCOM Workshops; date: 2018 | URL: https://ieeexplore.ieee.org/document/8406896/ . Earlier IEEE paper showing deep learning on 60K–260K Instagram and Carousell images can detect counterfeit sellers via shared images alone — the seed of the multi-modal seller-mapping literature.

3. Karunanayake, Rajasegaran, Gunathillake, Seneviratne & Jourjon, “A Multi-modal Neural Embeddings Approach for Detecting Mobile Counterfeit Apps: A Case Study on Google Play Store” | Publisher: arXiv preprint (extension of WWW 2019); date: 2020 | URL: https://arxiv.org/pdf/2006.02231 . Identified 2,040 likely-counterfeit apps among 49,608 lookalikes for the top 10,000 Google Play apps, and 1,565 with extra dangerous permissions — concrete evidence that multi-modal deep learning at platform scale uncovers infringement at orders of magnitude beyond manual review.

4. MarqVision, “Atomic Product Detections™” / “10 Best Brand Protection Tools Reviewed (2026)” | Publisher: MarqVision (vendor); date: 2026 | URLs: https://www.marqvision.com/anti-counterfeit-solution and https://www.marqvision.com/blog/brand-protection-tools . Industry vendor documentation of AI-driven counterfeit detection with claimed 99% accuracy across 1,500 marketplaces — illustrates that vendors now routinely ingest multi-modal data (images, text, video, social) at scale.

5. Red Points, “6 Best MarqVision Alternatives and Competitors in 2026” | Publisher: Red Points (vendor); date: 2026 | URL: https://www.redpoints.com/blog/best-marqvision-alternatives-and-competitors/ . Industry survey of major AI brand-protection vendors (Red Points, Corsearch, MarqVision, BrandShield, Bolster, CSC, OpSec) — useful as a market-level snapshot of the AI-enabled brand-protection vendor landscape.

B. Scale of infringement on platforms

6. OECD/EUIPO, “Mapping Global Trade in Fakes 2025: Global Trends and Enforcement Challenges” | Publisher: OECD/EUIPO; date: May 2025 | URL: https://www.oecd.org/en/publications/mapping-global-trade-in-fakes-2025_94d3b29f-en.html . Definitive recent estimate: counterfeit and pirated goods accounted for up to 2.3% of global trade ($467 billion) in 2021, and up to 4.7% of EU imports; small parcels (<10 items) jumped from 61% of seizures (2017–19) to 79% (2020–21), reflecting e-commerce-driven distribution. https://www.euipo.europa.eu/en/news/observatory/euipo-and-oecd-publish-a-report-on-counterfeit-and-pirated-trade

7. EUIPO & DG TAXUD, “EU Enforcement of Intellectual Property Rights: Results at the EU Border and in the EU Internal Market 2024” | Publisher: European Union Intellectual Property Office and DG TAXUD; date: October 2025 | URL: https://www.euipo.europa.eu/en/publications/eu-enforcement-of-intellectual-property-rights-results-2024 . 112 million counterfeit items detained in 2024 with retail value of €3.8 billion (vs. 152M and €3.4B in 2023); EUTM trademark infringements >84% of detained articles. Demonstrates the scale of detected infringement. https://www.euipo.europa.eu/en/news/european-union-seizes-112-million-counterfeit-items-worth-euro3-8-billion-in-2024 , https://www.euipo.europa.eu/en/publications/eu-enforcement-of-intellectual-property-rights-2024 , https://euipo.europa.eu/tunnel-web/secure/webdav/guest/document_library/observatory/documents/reports/2024_EU_Detentions/2024_EU_Enforcement_of_IPRs_FullR_en.pdf

8. U.S. Government Accountability Office, “Intellectual Property: Agencies Can Improve Efforts to Address Risks Posed by Changing Counterfeits Market,” GAO-18-216 | Publisher: U.S. GAO; date: January 2018 | URL: https://www.gao.gov/products/GAO-18-216 . Landmark GAO undercover study: 20 of 47 (≈43%) products purchased from third-party sellers on Amazon, Walmart, eBay, Sears, and Newegg were counterfeit; remains the most-cited primary U.S. evidence of marketplace counterfeit prevalence. https://www.cbsnews.com/news/amazon-walmart-newegg-third-party-sellers-sell-counterfeits-report-gao/

9. U.S. Department of Homeland Security, “Combating Trafficking in Counterfeit and Pirated Goods” | Publisher: DHS Office of Strategy, Policy, and Plans; date: January 2020 | URL: https://www.dhs.gov/sites/default/files/publications/20_0124_plcy_counterfeit-pirated-goods-report_01.pdf . DHS report on counterfeit trafficking via online platforms and recommendations to platforms; primary U.S. policy document.

10. Office of the U.S. Trade Representative, “2024 Review of Notorious Markets for Counterfeiting and Piracy” | Publisher: USTR; date: January 2025 | hosted on USTR docket: https://downloads.regulations.gov/USTR-2024-0013-0008/attachment_1.pdf . Annual U.S. government list naming Pinduoduo, DHGate, Taobao, Douyin Mall, and other major platforms as notorious markets; supports the thesis that major platforms host substantial counterfeit volumes. https://www.yahoo.com/news/articles/non-profit-calls-ustr-shein-130000271.html

11. ITIF, “How Chinese Online Marketplaces Fuel Counterfeits” / Comments to USTR for the 2025 Notorious Markets List | Publisher: Information Technology and Innovation Foundation; dates: Aug. 20, 2025 and Sept. 23, 2025 | URLs: https://itif.org/publications/2025/08/20/itif-investigation-finds-chinese-e-commerce-sites-facilitate-counterfeits/ and https://itif.org/publications/2025/09/23/comments-ustr-regarding-2025-review-notorious-markets-counterfeiting-piracy/ . ITIF found 25 of 43 successfully-delivered test purchases from Temu, AliExpress, and SHEIN were confirmed or likely counterfeits; recommends classifying all three as Notorious Markets.

C. Platform reluctance / safe-harbour-style structural friction

12. Tiffany (NJ) Inc. v. eBay Inc., 600 F.3d 93 (2d Cir. 2010) | Publisher: U.S. Court of Appeals for the Second Circuit | case overview: https://en.wikipedia.org/wiki/Tiffany_(NJ)_Inc._v._eBay_Inc. ; Finnegan analysis: https://www.finnegan.com/en/insights/articles/the-second-circuit-s-decision-in-tiffany-v-ebay.html . Foundational doctrine: platforms are not contributorily liable for trademark infringement absent specific knowledge — even though Tiffany’s test-purchase program found 73.1% of “Tiffany” goods on eBay were counterfeit. The structural baseline that lets platforms tolerate substantial counterfeit volume. https://cyber.harvard.edu/people/tfisher/IP/2010%20Tiffany%20Abridged.pdf , https://en.wikipedia.org/wiki/Tiffany_(NJ)_Inc._v._eBay_Inc.

13. Omega SA v. 375 Canal, LLC (2d Cir. 2021) | Publisher: K&L Gates analysis; date: January 19, 2021 | URL: https://www.klgates.com/Manufacturers-Must-Not-be-Blind-to-Their-Rights-Against-Counterfeiters-1-19-2021 . Reaffirms that “willful blindness” can yield contributory liability — but the threshold is high, and the doctrine continues to frame platform incentives.

14. Senator Coons & Tillis SHOP SAFE Act press releases (2021–2024) and SHOP SAFE Act of 2024 (House reintroduction by Issa & Nadler) | Publishers: U.S. Senate (Coons) / U.S. House; date: 2021–2024 | URLs: https://www.coons.senate.gov/news/press-releases/sens-coons-tillis-introduce-bipartisan-bicameral-shop-safe-actand https://natlawreview.com/article/shop-safe-act-e-commerce-trademark-enforcement-legislation-reintroduced-house . Documents the multi-year Congressional debate over imposing contributory trademark liability on platforms — and, by repeatedly failing to pass, illustrates the structural difficulty of changing platform incentives. https://www.sgrlaw.com/ttl-articles/shop-safe-act-a-bill-to-hold-e-commerce-sites-liable-for-counterfeit-goods-sold-online/

15. “Congress Is Trying to Keep Consumers Safe from Counterfeit Goods Online Through Legislation: Will It Work?” | Publisher: The Brand Protection Professional (Michigan State University); date: December 2023 | URL: https://bpp.msu.edu/magazine/congress-trying-consumers-safe-counterfeit-goods-december2023/ . Academic-trade analysis arguing the INFORM Consumers Act (effective June 2023) and SHOP SAFE proposals are necessary because existing doctrine and platform takedown procedures are “wildly ineffective” at curbing repeat infringement.

16. European Commission, “Commission Preliminarily Finds Temu in Breach of the Digital Services Act in Relation to Illegal Products on its Platform” | Publisher: European Commission; date: July 28, 2025 | URL: https://digital-strategy.ec.europa.eu/en/news/commission-preliminarily-finds-temu-breach-digital-services-act-relation-illegal-products-its . Preliminary EU finding that Temu’s risk assessment was “inaccurate” and consumers face “high risk” of encountering illegal products; possible 6%-of-global-turnover fine. Direct regulatory primary source on platform-level enforcement gaps. https://www.euronews.com/next/2025/07/28/eu-commission-finds-temu-in-breach-of-online-platform-rules , https://digital-strategy.ec.europa.eu/en/news/commission-preliminarily-finds-temu-breach-digital-services-act-relation-illegal-products-its , https://www.euronews.com/next/2025/07/28/eu-commission-finds-temu-in-breach-of-online-platform-rules

17. European Commission, formal proceedings against Shein (Feb. 2025) and Temu (Oct. 2024) under the DSA | Publisher: European Commission | URLs: https://digital-strategy.ec.europa.eu/en/news/commission-requests-information-shein-illegal-products-and-its-recommender-system and https://ec.europa.eu/commission/presscorner/detail/en/ip_24_5622 . Documents EU regulators’ active investigations targeting illegal/counterfeit-product dissemination by the largest emerging marketplaces.

18. “Counterfeit Crimes Unit: How Amazon fights counterfeits” | Publisher: Amazon (corporate news); date: 2025 retrospective | URL: https://www.aboutamazon.com/news/policy-news-views/amazon-counterfeit-crimes-unit . In five years, the CCU has filed 200+ civil actions and obtained $180M in court-ordered restitution; Amazon disposed of 15 million counterfeit products in 2024 alone — illustrating both the scale of the platform problem and the platform-controlled nature of enforcement. https://www.pymnts.com/amazon/2025/amazon-announces-3-actions-taken-against-counterfeiters/

[The post is available in full here, due to post length constraints of Beehiive]