Deepfakes—An Employment Law Perspective From Germany

00:00

24:53

Quick Hits

• Under Section 241 (2) of the German Civil Code (Bürgerliches Gesetzbuch (BGB)), the employer is obligated to respond promptly upon becoming aware of deepfake incidents in the workplace; failure to do so may expose the employer to substantial claims for damages and compensation for pain and suffering.
• The creation and distribution of deepfakes with sexual or degrading content constitutes a serious breach of the duty of mutual consideration, which may regularly justify extraordinary termination pursuant to Section 626 (1) BGB.

In the workplace context, various scenarios are conceivable: the creation of deepfakes with sexual or degrading content depicting colleagues (whether using company or private IT infrastructure), deepfakes targeting supervisors to damage their reputation, and the use of deepfakes to manipulate business operations—for example, through forged instructions in audio or video form.

For those affected, such content can have severe personal and professional consequences—ranging from psychological distress to reputational harm to disruption of their daily work. For employers, the question arises as to how they can, on the one hand, fulfill their duty of care toward affected employees and, on the other hand, address the (alleged) perpetrators under employment law.

Employer’s Duty of Care

By virtue of the duty of care arising from Section 241 (2) BGB in conjunction with Articles 1 (1) and 2 (1) of the German Basic Law (Grundgesetz (GG)), the employer is obligated to protect the personal rights (Persönlichkeitsrechte) of its employees. If an employee becomes the victim of deepfakes and a workplace nexus exists, the employer must take active protective measures.

The principles developed by case law in the context of workplace bullying (Federal Labor Court (Bundesarbeitsgericht (BAG)), May 16, 2007, case no. 8 AZR 709/06) are likely transferable to this context. According to these principles, the employer has, among other things, an organizational duty to structure the workplace in a manner that safeguards personal rights. The employer is already obligated to act when it becomes aware—or should have become aware—of concrete indications of violations of personal rights. As with workplace bullying, the creation and distribution of deepfakes constitutes a serious violation of personal rights that can cause significant psychological distress.

Once the employer becomes aware of deepfake incidents, it must respond promptly. Measures to be considered include preventing further distribution (e.g., by blocking company IT systems), separating the perpetrator from the victim (e.g., through reassignment or paid leave), and imposing employment law sanctions up to and including extraordinary termination of the (suspected) perpetrator.

If the employer fails to take appropriate measures, it may face substantial claims for damages and compensation for pain and suffering from the victim. In addition, the affected employee may, for his or her part, terminate the employment relationship without notice pursuant to Section 626 BGB.

Furthermore, employers may want to take preventive action by, for example, explicitly addressing the use of AI in (ethics) guidelines. In this context, it is also important to consider the employer’s liability under the EU AI Act for the AI tools provided at the workplace.

Extraordinary Termination of the Perpetrator

Where it is established that an employee has created or distributed deepfake content depicting colleagues, extraordinary termination pursuant to Section 626 (1) BGB may be considered (cf. on bullying cases: Thuringia State Labor Court (LAG), February 15, 2001, case no. 5 Sa 102/00). The creation of deepfakes—particularly those with sexual or degrading content—constitutes a violation of the affected colleague’s personal rights and thereby a serious breach of the duty of mutual consideration under Section 241 (2) BGB, which—depending on its severity—may irreparably destroy the relationship of trust and sustainably disrupt workplace harmony.

In the balancing of interests—which is required before any termination is pronounced—the employer’s interest in terminating the employment relationship will generally prevail in cases of sexually motivated deepfakes, since the personal rights of those affected are violated at their core and the employer would otherwise breach its duty of care. The specific circumstances of the individual case—such as length of service, prior conduct, the nature of the content, and the extent of its distribution—must nonetheless be considered.

Where the identity of the perpetrator cannot be established beyond doubt, a termination based on suspicion may be considered. According to the settled case law of the Federal Labor Court, this requires that strong grounds for suspicion exist that are based on objective facts, that the employer has undertaken all reasonable investigative measures, and that the suspected employee has been given a proper hearing beforehand in order to present any exculpatory circumstances (BAG, November 29, 2007, case no. 2 AZR 724/06).

Particularly in the case of deepfakes, attributing the content to a specific employee frequently poses difficulties, as creation typically occurs on private devices. The employer may want to therefore carefully document the investigation and preserve IT forensic findings. In the case of a termination based on suspicion, the two-week filing period under Section 626 (2) BGB does not begin to run until the completion of all reasonable internal investigations, including the hearing of the employee, although the employer is generally required to proceed with due diligence and without undue delay.

Requirement of a Prior Warning

As a general principle, a conduct-based termination must be preceded by a formal warning in order to give the employee an opportunity to change his or her behavior. According to the case law of the Federal Labor Court, however, a warning is dispensable where the breach of duty is so serious that the employee could not have expected the employer to tolerate such conduct from the outset (BAG, May 20, 2021, case no. 2 AZR 596/20).

Where the allegation involves the creation and/or distribution of deepfakes depicting colleagues with sexual or degrading content, it appears justifiable to consider a prior warning dispensable. The creation of such content violates the core of the contractual duties of mutual consideration. Every employee must understand that such behavior will not be tolerated by the employer. Even if the creation or distribution occurs during the employee’s personal time, the workplace nexus and the associated potential harm should be readily apparent.

A different assessment may apply in borderline cases, such as manipulated content without a sexual dimension that is more in the nature of a tasteless joke. In such cases, a formal warning may come into consideration as a less severe measure, with the distinction depending in each individual case on the nature and severity of the content, the extent of distribution, and the impact on those affected.

In any event, workplace guidelines on AI use are likely to prove helpful at this juncture, ideally specifying which conduct may result in employment law consequences.

Pressure Termination

Where the workforce pushes for the dismissal of the perpetrator, the question arises as to the permissibility of a so-called pressure termination. However, this judicially recognized concept is subject to strict requirements. The Federal Labor Court distinguishes two categories of cases: where the pressure is based on an objectively justified ground for termination, the dismissal may already be effective as a standard conduct-related or person-related termination. Where no independent ground for termination exists, the employer must first protect the employee (alleged perpetrator) and take all reasonable measures to resist the pressure. Only if these measures fail and the threatened disadvantages for the employer render continuation of the employment relationship unreasonable may a termination for operational reasons be considered (BAG, October 4, 1990, case no. 2 AZR 201/90).

Where the creation of deepfakes has been proven, an independent ground for termination will generally already exist, such that the principles governing pressure terminations in the strict sense are not dispositive. The pressure from the workforce then merely serves as an additional indicator of a sustained disruption of workplace harmony.

The pressure termination becomes practically relevant above all where the identity of the perpetrator cannot be clearly established, but the suspicion is known within the workplace. In such cases, the employer must first de-escalate the situation through internal measures—such as reassignment or paid leave of the suspect—while at the same time safeguarding the rights of the accused. Only where all reasonable measures have been exhausted and the pressure reaches a level that seriously jeopardizes business operations (e.g., collective refusal to work) may a pressure termination be considered as a last and proportionate resort. However, the legal requirements for such a so-called ‘real’ pressure termination are high.

Conclusion

The employment law challenges posed by deepfakes will grow in significance as the corresponding AI tools become increasingly accessible. Employers may want to establish clear codes of conduct early on and to respond decisively to incidents. Existing case law on workplace bullying and sexual harassment provides helpful points of reference, even though the specific challenges of deepfake technology—particularly with regard to fact-finding and perpetrator identification—will require further development of existing legal principles.

Ogletree Deakins’ Berlin and Munich offices will monitor developments and post updates on the Germany and Technology blogs as additional information becomes available.

Tatjana Serbina is counsel in the Berlin office of Ogletree Deakins.

Pauline von Stechow, a law clerk in the Berlin office of Ogletree Deakins, contributed to this article.

Follow and Subscribe
LinkedIn | Instagram | Webinars | Podcasts