Deepfakes: Majority Expect to Be Deceived by AI Manipulation

• 55 per cent believe they are likely to be deceived by deepfakes or other AI manipulation within the next five years  

• 28 per cent expect damaging or abusive AI content to be created about them  

• eco calls for clear rules, understandable labelling and better awareness-raising 

Berlin, 5 August 2026 – Fifty-five per cent of adults in Germany believe it is very or fairly likely that they will be deceived by AI-generated deepfakes or similar AI manipulation within the next five years, for example in connection with fraud or information theft. By contrast, only 28 per cent believe it is likely that damaging or abusive AI content will be created about them. 

These are the findings of a representative YouGov survey commissioned by eco – Association of the Internet Industry. A total of 2,012 people aged 18 and over in Germany were surveyed between 31 July and 2 August 2026. 

Respondents therefore consider the likelihood of being deceived by manipulated content to be almost twice as high as the likelihood of becoming the subject of such manipulation themselves. 

Deepfakes are image, audio or video content created or altered using artificial intelligence that can appear authentic. They may be used for identity misuse, attempted fraud, reputational damage, sexualised digital violence, as in the Collin Fernandes case, or disinformation, among other purposes. 

Deepfakes cause harm in two ways: people can be harmed by manipulated content about them. But they can also become victims because they believe a fake to be genuine and act on that basis,” says Alexandra Koch-Skiba, Head of the eco Complaints Office. “The fact that 55 per cent consider such deception likely shows just how much pressure trust in digital content is under. These figures are a societal warning signal.

Labelling alone does not prevent deception

The transparency obligations under Article 50 of the European AI Act have applied since 2 August 2026. Providers must label certain AI-generated or manipulated content in a machine-readable format. Deepfakes must also be disclosed as having been artificially generated or manipulated. 

“Labelling is an important building block, but it is not a shield against fraud, identity misuse or digital violence,”, says Koch-Skiba. “Alongside reliable technical standards and clear notices, we need greater media literacy and effective reporting infrastructure. People must be able to recognise attempted manipulation and verify suspicious content through a second communication channel.” 

Tackling harmful deepfakes requires clear rules and a range of measures 

eco welcomes the objective of the planned Act to Strengthen Civil and Criminal Law Protection Against Digital Violence (GgdG). Those affected must be able to report unlawful deepfakes quickly, preserve evidence and enforce their rights to information and deletion. 

At the same time, the Association is calling for clear legal terminology, judicial oversight and close alignment with the European Digital Services Act. Unclear liability rules must not lead platforms to remove lawful artistic, satirical or journalistic content as a precaution. 

In addition to consistent law enforcement, investment in prevention and digital education is needed. Criminal-law measures generally take effect only after harm has already occurred. 

“Protection against deepfakes is not determined in court alone,” Koch-Skiba emphasises. “It begins with awareness-raising, digital education and the responsible handling of AI-generated content.” 

This position is consistent with eco’s statement on the draft bill (in German). 

About the eco Complaints Office 

The eco Complaints Office has been active since 1996 and is an integral part of Germany’s system of regulated self-regulation. Sie ermöglicht Internetnutzerinnen und Internetnutzern die kostenlose und auf Wunsch anonyme Meldung jugendgefährdender und strafbarer Inhalte und arbeitet eng mit Diensteanbietern, Behörden und internationalen Partnerorganisationen zusammen. Hierbei wird die schnelle Löschung rechtswidrige 

• The latest annual report is available to download here. 

• The background paper marking 30 years of the eco Complaints Office is available here. (in German) 

About eco 

With around 1,000 member companies, eco is the largest Internet industry association in Europe. Since 1995, eco has played a decisive role in shaping the Internet, promoting new technologies, developing framework conditions and representing its members’ interests in dealings with policymakers and in international bodies. Its key areas of focus are the reliability and strengthening of digital infrastructure, IT security and trust, and ethics and self-regulation. 

Survey methodology 

The data for this survey are based on online interviews with members of the YouGov panel who had agreed in advance to participate. A total of 2,012 people were surveyed between 31 July and 2 August 2026. Quotas were applied for age, gender and region, and the results were subsequently weighted accordingly. The results are representative of Germany’s resident population aged 18 and over.