Enforcement of the AI Act: New Risks for Companies and Creators

With core provisions of the AI Act (EU Regulation, KI-VO; hereinafter, the “AI Act”) becoming applicable as of 2 August 2026, the competition-law enforcement of its requirements is also set to gain increasing importance. Going forward, infringements of the AI Act could not only be sanctioned by regulators but could equally become the subject of competition-law cease-and-desist notices (Abmahnungen). What deepfakes actually are and why they are relevant under competition law can be read in more detail here. Against this background, the key question is under what conditions extensive cease-and-desist activity can still be regarded as legitimate enforcement of rights, and at what point it must instead be regarded as abusive conduct.

I. Identifying Abusive Cease-and-Desist Notices

As a general rule, extensive cease-and-desist activity alone is not sufficient to establish an abuse of rights, provided that numerous competition-law infringements are genuinely at issue at the same time. Where many competitors are infringing competition-law provisions, a company must have the ability to pursue these infringements by way of multiple cease-and-desist notices.

A finding of abuse of rights requires additional circumstances indicating that the cease-and-desist notices do not primarily serve the fairness of competition, but instead serve extraneous motives that constitute the actual driving force behind them. After all, the pursuit of competition-law infringements is, in principle, also in the interest of other market participants and of the public at large.

That said, even a single cease-and-desist notice can amount to an abuse of rights where there are sufficient indications that it predominantly pursues objectives other than the protection of competition. Such abuse is particularly likely where the party issuing the notice can barely be said to have any material economic interest in pursuing the competition-law infringement, yet nevertheless instructs counsel to issue a large number of cease-and-desist notices. This applies with particular force where, owing to limited financial capacity, that party could hardly bear the cost risk associated with the notices itself. An abusive enforcement of rights is further indicated where the scope and intensity of the cease-and-desist activity bear no reasonable relationship to the issuing party’s own actual commercial activity.

The law of unfair competition is specifically not intended to give companies the opportunity to act as general “guardians of competition,” independent of their own economic interests.

II. Why Deepfakes Could Provide a Basis for (Abusive) Mass Cease-and-Desist Campaigns

A significant risk of standardized cease-and-desist activity could develop in the future, particularly in relation to AI-generated content. This is because, similar to what has already been observed in e-commerce in relation to imprint, price-disclosure and information obligations, infringements of the labeling obligations imposed by the AI Act typically involve facts that can often be identified online without any great effort. Whether or not publicly accessible content has been labeled as AI-generated can regularly be established simply through a visual inspection of social media posts, websites, or advertising content.

A particular source of conflict is that the scope of the term “deepfake” does not yet appear to be conclusively settled. While the term is frequently understood to refer to deceptively realistic image or video content, a broader understanding of the concept is increasingly emerging. For example, the Wettbewerbszentrale (German Centre for Protection against Unfair Competition), in its guidance issued in February 2026, likewise emphasized that the labeling obligations are unlikely to be limited to fully photorealistic content alone. Rather, the obligations could already capture depictions that merely contain individual realistic elements or that are capable of creating an authentic impression among parts of the public.

It is precisely this lack of clarity that is likely to considerably increase the potential for cease-and-desist notices. The more broadly the term “deepfake” is construed, the greater the number of content items potentially subject to a labeling obligation. Those affected could therefore include not only professional media companies or major platform operators, but in particular also influencers, agencies, online retailers, small businesses, and content creators who use AI-generated content in advertising or public-facing communications.

At the same time, this creates structures typical of those already familiar from abusive mass cease-and-desist campaigns in digital commerce: easily identifiable online infringements, standardized review mechanisms, and a large number of potential respondents.

The situation would be even further-reaching if the labeling obligations became relevant not only for content published in the future, but also for deepfakes already uploaded in the past. Such an interpretation does not appear far-fetched. For content that remains publicly accessible, it could be argued that the infringement lies not solely in the original upload, but continues in the ongoing provision of an unlabeled deepfake in its current form. Should such an understanding prevail, the potential for cease-and-desist notices would expand considerably, since not only new publications but potentially also extensive existing content libraries could subsequently be reviewed and challenged.

III. Response Options and Defense Strategies Against Abusive Cease-and-Desist Notices

Where there are indications that a cease-and-desist notice is abusive, the recipient should not hastily comply with the claims asserted. Rather, it is advisable first to have the notice carefully reviewed as to whether it is justified at all. This is because an abusive cease-and-desist notice generally gives rise neither to a claim for reimbursement of the costs of the notice nor to any further-reaching claims arising from the cease-and-desist relationship. Nor, as a rule, do obligations to provide information or to respond arise in response to a factually unjustified notice.

A party wrongly subjected to a cease-and-desist notice is, moreover, not confined to a purely defensive response, but may itself take active steps against the abusive enforcement of rights. In particular, a counter cease-and-desist notice (Gegenabmahnung) may be considered. By means of such a counter notice, the recipient can, in turn, call upon the purported claimant to cease and desist, or can announce the filing of a negative declaratory action. Such a counter notice can be especially useful where it is to be expected that the party issuing the original notice will abandon its claims once the facts have been supplemented or the legal position clarified. This may be the case, for example, where the original notice was based on inaccurate facts, where material factual or legal aspects were not previously taken into account, or where the issuing party could not have recognized the recipient’s lack of responsibility for the alleged infringement.

A counter cease-and-desist notice can also be significant from a procedural perspective. If a negative declaratory action is filed immediately, without first giving the opposing party the opportunity to withdraw its claims, there is a risk of being ordered to bear the costs under Section 93 ZPO (German Code of Civil Procedure) if the opposing party immediately acknowledges the claim.

In addition, Section 13(5) UWG (German Unfair Competition Act) now expressly provides that a party wrongly subjected to a cease-and-desist notice may claim reimbursement of the expenses necessarily incurred in defending itself. While this does not automatically cover the full costs of a counter cease-and-desist notice, there are today considerably better options for defending oneself, including on economic terms, against unjustified cease-and-desist notices.

If an abusive cease-and-desist notice has already led to the conclusion of a cease-and-desist agreement (Unterlassungsvertrag), termination of the agreement for good cause under Section 314 BGB (German Civil Code) may also be considered. In individual cases, rescission on grounds of fraudulent misrepresentation under Section 123(1) BGB may also be possible. If the party that issued the notice demands payment of a contractual penalty despite the abuse of rights, this demand can generally be countered with the defense of abuse of rights. In addition, the recipient of the notice may, depending on the circumstances, also have its own claims for damages.

IV. Conclusion

With the AI Act’s new labeling obligations, deepfakes could become a significant competition-law risk. In particular, the ease with which potential infringements can be identified, the as yet unclear scope of the term “deepfake,” and the large number of potentially affected content items create risks of abuse. Should previously published content also come to be covered in the future, a considerable expansion of potential cease-and-desist activity looms. Companies, influencers, and content creators are therefore well advised to review AI-generated content and existing content libraries at an early stage for possible labeling obligations.