New Proactive Measures

As platforms move to supplement notice and takedown tools with more proactive measures, they are relying heavily on technology to power monitoring.

Central and South American online marketplace Mercado Libre has set up a system to detect IP-infringing listings, without the need for a brand owner’s report.

“Once we detect a suspicious seller, various ratios are calculated so that we can ban them. Furthermore, we ban all the rest of the duplicate accounts that they might have,” explained Juan Cichero, Head of Brand Protection–Marketplace, at Mercado Libre (Argentina).

“After that, we make our re-registration process harder for all sellers who have strong relations with accounts that had been banned before.”

The process involves machine learning to identify infringing listings and automatically remove them. The machine learning models mainly “learn” from the reports submitted by brand owners through the reporting tool.

“The more brand owners use our tool to monitor and report, the more our models will proactively detect and remove allegedly infringing listings. In this way, the success of our proactive measures mainly depends on the efforts of the brand owners,” Mr. Cichero said.

Alibaba describes its proactive monitoring as one of its most powerful enforcement mechanisms.

“Given the scale of Alibaba’s e-commerce platforms, it would not be possible to take effective proactive action without cutting-edge technology,” Mr. Bassiur said.

Each year, he said, the number of listings proactively removed by Alibaba “is many times greater than the number of listings removed as a result of rights holder requests.”

Alibaba’s proactive screening algorithms incorporate product intelligence, real-time monitoring and interception, biometric identification, image and semantic recognition algorithms, and algorithms to detect anomalous traffic and merchant behaviors.

For example, in order to monitor sales behavior, Alibaba uses algorithms to search for anomalous traffic data and unusual transactions which, when identified, trigger deeper analysis technologies and additional investigation.

It uses this to detect potentially problematic activity, such as abrupt changes from one category of products to another, or periodic or unusual timing of sales.

The platform has also introduced live broadcasting controls—a growing marketing tool that allows merchants to connect with consumers, provide product demonstrations to their followers, and share product information in real time.

Alibaba uses voice and semantic recognition algorithms and image and character recognition algorithms to monitor the live broadcasting process. If counterfeiting activity is identified, the merchant is subjected to penalties.

Amazon’s Active Protection

Amazon has designed three tools that enable brands to help protect their IP on Amazon’s online stores.

“We operate on a presumption that simply being reactive in a notice-and-takedown sense is not enough to protect IP, so we invest very heavily in proactive measures to prevent infringing material from appearing on our store in the first place,” said Charles Wright, Associate General Counsel at Amazon.com (US).

The first tool is Brand Registry, a free service that enables a trademark owner, in any of the countries Amazon operates in, to alert Amazon of its trademarks.

“That allows us to proactively protect those trademarks in our catalogue by doing things such as scanning for misuse of the trademarks, whether it’s a word or logo mark,” Mr. Wright explained.

The platform can place restrictions on who can sell products using the trademark if it knows there are already restrictions based on geography and where those products are distributed, he added.

Brand Registry includes an IP infringement-reporting tool, called Report a Violation. When a rights owner reports a violation through this tool, the information is fed into Amazon’s machine learning model, which explores why the violation was missed by the automated tool in the first place.

“It asks: ‘What is it about this violation that caused us to miss it?’ and thereby improves our proactive protection,” Mr. Wright said.

“While we shoot for an end goal where brands would never have to provide us with notices of violation, it remains very important that brands report things so we can learn what we are missing,” he added.

Another tool used by Amazon is Transparency, whereby brand owners inscribe each individual item with a unique QR code. For example, with iPhones, Apple generates a code during the manufacturing process that is unique to each phone.

“This unit-level serialization allows us and other retailers to ask whether a particular product should have a code on it. If it doesn’t have one, we know it’s not an authentic product,” Mr. Wright explained. “If it does have one, we can scan the code to confirm that the code was issued for that item.”

Amazon’s third solution, Project Zero, is a self-service removal tool. Amazon flips the presumption of notice and takedown for trusted partners who, it believes, are not going to abuse its tools and will use them solely for IP protection. Under this system, trust matters—users can remove infringing listings directly from the platform.

“Just like with our notice forms, when these brand owners make removals, we use that information again to ask: ‘Why did we miss this in the first place?’,” Mr. Wright said.

“We have one rule with Project Zero: if you abuse it, you lose it. Unfortunately, we have had to terminate access for several brands who were using it not to protect their IP, but instead to try to police their distribution agreements.

“Those are two very different things. The sale of infringing goods and counterfeit goods is illegal; the sale of authentic goods by someone not contracted to sell them is a matter of contract, but not illegal,” he said. “We don’t want people attempting to police their distribution agreements using our IP tools.”

In June this year, the platform launched a Counterfeit Crimes Unit, a global team with specialized experience in investigating and bringing legal action against bad actors. The team consists of former federal prosecutors, experienced investigators, and data analysts.

Amazon is also taking its anticounterfeiting fight to the courts. The platform has partnered with a number of companies, from luxury brands to family businesses, to file joint lawsuits against counterfeiters.

On November 12, Amazon filed a lawsuit against two influencers and 11 third-party merchants at the U.S. District Court for the Western District of Washington, for allegedly advertising, promoting, and facilitating the sale of counterfeit luxury goods in Amazon’s store. The influencers and sellers allegedly conspired to evade Amazon’s anticounterfeiting protections by promoting counterfeit products on Instagram and TikTok as well as their own websites.

At the time, Cristina Posa, Associate General Counsel and Director, Amazon Counterfeit Crimes Unit (US), in a company press release, said: “This case demonstrates the need for cross-industry collaboration in order to drive counterfeiters out of business. Amazon continues to invest tremendous resources to stop bad actors before they enter our store and social media sites must similarly vet, monitor, and take action on bad actors that are using their services to facilitate illegal behavior.”

Facebook Fights the Fakes

Mark Fiore, Director and Associate General Counsel at Facebook, Inc. (US), said the company’s primary focus, and a top priority over the last couple of years, has been instituting screening measures to prevent counterfeits—also a proactive approach.

Previously, Facebook’s work was focused on reactive measures, such as its notice and takedown program.

“That work continues to be very important for us, and we have made a number of enhancements over the years, including moving from weekday coverage for processing IP reports to having it 365 days a year, 24/7.

“All of this has resulted in a quicker turnaround time,” Mr. Fiore said, noting that Facebook now typically responds to reports within 24 hours, and often within a few hours or even minutes.

With its proactive approach, the platform primarily focuses on two areas: how it handles situations when it receives a report from a rights owner, and blocking content that is potentially infringing before, or without, it being reported by a rights owner.

Sabrina Perelman, Associate General Counsel at Facebook (US), said the platform regularly takes down more content than a rights owner has specifically reported.

“Facebook views this as ‘enhanced report processing,’ or simply connecting the dots between the specific reported content and actions Facebook itself might be able to take,” Ms. Perelman said.

For example, if a rights owner reports a single or multiple posts or listings on Facebook or its sister platform Instagram, Facebook will, in addition to taking down those specific links, look at the larger context in which it was posted.

This means reviewing the entire page, Instagram account, or Facebook group, to see whether there is evidence of widespread, similar activity.

“If so, we may be able to take down the entire actor, even based on a single report,” she said.

Additionally, when a rights owner reports an advertisement related to counterfeits, Facebook will take down not only the advertisement but also the account that placed it, and it will block financial instruments that were used to place ads, bank accounts, and credit cards.

“That has the automatic effect of taking down any related accounts tied to the same financials, with the stroke of a key based on just one report,” Ms. Perelman explained.

Facebook also employs a repeat infringer policy. “If for some reason we are not able to take action right away from the first report, we count IP strikes behind the scenes and will ultimately remove actors if they are repeatedly flagged for IP infringement,” she added.

As part of its proactive measures taken before, or without, a rights owner’s report, Facebook reviews advertising marketplace listings before they go live for alignment with its policies which explicitly prohibit IP infringement.

It uses signals such as combinations of certain brand names and keywords, that may indicate when a counterfeit is being sold. Other key signals are unrealistically low prices or signs of nefarious activity.

In addition, Facebook has measures in place on Instagram to block hashtags that lead users to infringing content.

“If you search for a hashtag such as #(brandname)replica, you should receive zero results,” Ms. Perelman said.