Brands Under Attack: How Artificial Intelligence Has Supercharged Counterfeiting, and What Companies Can Do About It

Counterfeit products have never been a novelty in Brazil. Copied clothing, electronics, medicines and cosmetics have circulated for decades in street markets and, more recently, on digital marketplaces. What has changed in recent years is the scale and sophistication of the problem, driven by generative artificial intelligence tools. Today it is possible to create packaging that is nearly identical to the original, generate realistic photos of products that never physically existed, and even produce videos showing executives or influencers endorsing brands they have never worked with. Deepfakes, once associated mainly with political disinformation or non-consensual pornography, have also become a tool for commercial fraud.

For companies that spent years building a reputation and went through the effort of registering their trademarks, this scenario poses a direct risk to revenue, consumer trust and corporate identity. And the problem is no longer theoretical: reports of fake social media ads using AI-generated faces to sell fraudulent products, or manipulated videos attributing statements to company executives, are already a documented reality across several sectors, from retail to finance.

What Brazilian law already covers, and where it still falls short

The Industrial Property Law, Law 9,279 of 1996, remains the main legal tool against counterfeiting. It criminalizes the reproduction or imitation of a registered trademark in a way that causes confusion, and also provides for the search and seizure of counterfeit products and compensation for damages. Registering a trademark with the National Institute of Industrial Property, the INPI, continues to be the first and most important step for a company to have standing to take legal action against anyone copying it, whether on physical packaging or in digital ads.

The challenge arises when counterfeiting stops being purely physical and becomes synthetic. An AI-generated image that reproduces a brand's visual identity without literally copying a registered logo may escape a superficial trademark infringement analysis, but it can still amount to unfair competition under article 195 of the Industrial Property Law itself, which addresses acts capable of creating confusion between products or business establishments. Brazilian case law, including rulings from the Superior Court of Justice (STJ), has already established that trademark protection goes beyond literal copying, covering any practice that leads consumers to err as to the origin of a product or service.

Deepfakes and the misuse of imagery tied to a brand

A second line of attack involves using the image of people, whether celebrities hired for campaigns or the company's own executives, in AI-manipulated videos. These deepfakes are often used to lend credibility to financial scams, sales of counterfeit products, or fraudulent investment promises, exploiting the trust the public places in that person or brand. Legally, this practice touches at least three areas of protection: the right to image and personality, guaranteed by article 20 of the Civil Code and by article 5, item X, of the Federal Constitution; trademark protection against unauthorized use that creates an improper association; and, in many cases, the criminal classification of fraud or electronic fraud, when there is actual financial harm to the deceived consumer.

For the company victimized by this kind of manipulation, the damage is not limited to the isolated incident. There is a diffuse reputational effect: consumers who were harmed by a scam associated with the brand tend to hold the company responsible, even when it is as much a victim as they are. This is a point where legal counsel needs to work alongside communications and compliance teams, because a legal response alone, without proper crisis management, is rarely enough to contain the damage to the brand.

Liability of digital platforms and the role of the Marco Civil

Much of the circulation of counterfeit products and fraudulent synthetic content happens on social media platforms and marketplaces. The Brazilian Internet Civil Framework (Marco Civil da Internet), Law 12,965 of 2014, establishes that, as a general rule, application providers are only liable for third-party content if, after a specific court order, they fail to take action to remove the material flagged as infringing. This requirement of a prior court order has been the subject of intense debate at the Federal Supreme Court, especially in recent rulings on the constitutionality of article 19 of the law, which discuss whether platforms should bear broader liability for clearly unlawful content, such as fraudulent deepfakes and openly advertised counterfeit products.

In practice, this means that companies that identify fake ads or manipulated videos using their brand should not settle for an informal takedown request. It is advisable to formally notify the platform, document all evidence with dates, URLs and screenshots, and, when removal does not happen voluntarily or proves insufficient, seek urgent injunctive relief from the courts to compel the content's removal and, where applicable, the identification of whoever published it.

LGPD and the use of data to train counterfeiting tools

Another point that deserves attention from companies is the source of the data used to train AI models capable of generating these counterfeits. Images from advertising campaigns, product photos, recordings of corporate events and even internal training videos can be collected without authorization and used to feed systems that later produce fraudulent content. The General Data Protection Law (LGPD), Law 13,709 of 2018, although not specifically designed to deal with generative AI, already provides a relevant legal basis: the processing of personal data, including image and voice, requires a specific purpose and, as a rule, consent or another appropriate legal basis. Companies that discover their institutional material was used without authorization to train tools that later generated fraudulent deepfakes can explore this avenue of liability, even though the subject still lacks more specific regulation on artificial intelligence in Brazil.

Bill 2338/2023, currently moving through Congress and aimed at establishing a general regulatory framework for artificial intelligence in the country, is the clearest sign that this regulatory gap is likely to narrow. The bill provides for classifying systems by risk level, transparency requirements when content is generated or manipulated by AI, and clearer civil liability mechanisms. Until it is definitively approved, however, companies will keep relying on a combination of existing rules, Industrial Property Law, the Civil Code, the Consumer Protection Code and the LGPD, interpreted together to address a problem that the legislation, taken in isolation, did not precisely anticipate when it was drafted.

Practical protective measures for companies

Given this scenario, a few preventive measures make a concrete difference in a company's ability to respond. Keeping trademark registration up to date and comprehensive, including logo variations and related product classes, is the foundation for any future action. Actively monitoring marketplaces and social media, whether with in-house tools or specialized services, allows counterfeits and fraudulent content to be identified before they gain significant reach. Systematically documenting every occurrence, recording the date, taking screenshots and, where possible, obtaining a notarized record (ata notarial), strengthens the company's position if litigation becomes necessary. And having a clear internal policy on the use of executives' and spokespeople's images in advertising campaigns helps establish a recognizable pattern for the public, making it easier to identify and report fake content.

Brand protection is no longer just a matter of formal registration. It now requires continuous monitoring and a fast response, since the length of time fraudulent content stays up is often what determines the extent of the damage. Companies that treat this issue in a strictly reactive way, acting only after the problem has already caused significant harm, tend to face longer and costlier processes to repair their image. For this reason, preventive legal analysis, covering everything from trademark registration strategy to the definition of digital incident response protocols, is increasingly becoming a core part of risk management for any business that depends on reputation to stay competitive.