We've been getting this question almost weekly from clients planning their social media content: "Can we still use AI-generated images on Instagram? And do we need to label them?" The background: starting 2 August 2026, new EU transparency requirements apply to certain AI-generated content. The deadline is real, and so is the uncertainty around it, but the answer is more nuanced than many assume at first glance.
What actually changes on 2 August 2026
On this date, core provisions from Article 50 of the EU AI Act come into force. This isn't a blanket requirement to label every AI-assisted text or every lightly edited image file. The focus is on realistic-looking content that could be mistaken for authentic material: AI-generated images and videos, synthetic voices, and certain automatically generated texts, particularly when made publicly available. A hotel room image created entirely with an AI tool, or an ad clip using a synthetic voice, clearly falls into this category. How heavily an edited photo needs to be altered before the requirement applies depends on the individual case.
Does this affect businesses in Switzerland too?
This is where most of the confusion comes from. Switzerland currently has no general, standalone labelling requirement for AI content. Existing law, covering data protection, personality rights, and copyright, among other things, continues to apply regardless.
The EU AI Act itself, however, has extraterritorial reach, similar to what many already know from GDPR. What matters isn't where a business is based, but where its content reaches its audience. As soon as a business reaches customers, guests, or an audience within the EU, the EU AI Act should be treated as a baseline standard. For most Swiss businesses with a website, a social media presence, or an international audience, that threshold is crossed more often than one might initially think.
This is especially clear in hospitality: any business hosting international guests, which describes most Swiss hotels or restaurants, should take the labelling requirement just as seriously as a hotel in Germany or Austria would. The effort involved is manageable; the risk of ignoring it is unnecessarily high.
What this means for social media content in practice
A simple framework helps in day-to-day work:
Clearly affected: fully AI-generated images (e.g. a room or product image created with Midjourney), AI-generated video clips, synthetic voices in ad clips, and heavily AI-altered photos that create a realistic impression.
Generally not affected: internal use of AI for text support, brainstorming, translations, or editorially reworked content where a human retains editorial responsibility and makes substantial changes.
The labelling itself doesn't need to be complicated. A clearly visible note that an image or video was AI-generated or AI-edited is sufficient in most cases. What matters is that it isn't hidden or buried in fine print.
Why this is an opportunity, not just an obligation
Interestingly, this regulation accelerates a trend that was already underway: the value of genuine, camera-produced content is rising. Once AI images have to carry a visible label, the distinction between "artificially generated" and "genuinely produced" becomes clearly visible to audiences for the first time, and more relevant to how a brand is perceived.
For businesses and hotels, this means: those who continue working with real photos and video don't need to worry about this requirement, and position themselves as authentic at the same time, in an environment that increasingly distinguishes between real and artificial content. That's not a coincidence; it's a shift worth positioning for early.
Conclusion
The AI labelling requirement starting in August 2026 isn't a reason to panic, but it is a good occasion to review your own content strategy: which content is currently created with AI tools, which with genuine camera production, and how transparently is that communicated to customers and guests? Clarifying these questions now, rather than improvising in August, saves stress, and strengthens credibility along the way.
This article does not constitute legal advice. For questions specific to your own situation, we recommend consulting a lawyer specialising in data protection and IT law.
We've been getting this question almost weekly from clients planning their social media content: "Can we still use AI-generated images on Instagram? And do we need to label them?" The background: starting 2 August 2026, new EU transparency requirements apply to certain AI-generated content. The deadline is real, and so is the uncertainty around it, but the answer is more nuanced than many assume at first glance.
What actually changes on 2 August 2026
On this date, core provisions from Article 50 of the EU AI Act come into force. This isn't a blanket requirement to label every AI-assisted text or every lightly edited image file. The focus is on realistic-looking content that could be mistaken for authentic material: AI-generated images and videos, synthetic voices, and certain automatically generated texts, particularly when made publicly available. A hotel room image created entirely with an AI tool, or an ad clip using a synthetic voice, clearly falls into this category. How heavily an edited photo needs to be altered before the requirement applies depends on the individual case.
Does this affect businesses in Switzerland too?
This is where most of the confusion comes from. Switzerland currently has no general, standalone labelling requirement for AI content. Existing law, covering data protection, personality rights, and copyright, among other things, continues to apply regardless.
The EU AI Act itself, however, has extraterritorial reach, similar to what many already know from GDPR. What matters isn't where a business is based, but where its content reaches its audience. As soon as a business reaches customers, guests, or an audience within the EU, the EU AI Act should be treated as a baseline standard. For most Swiss businesses with a website, a social media presence, or an international audience, that threshold is crossed more often than one might initially think.
This is especially clear in hospitality: any business hosting international guests, which describes most Swiss hotels or restaurants, should take the labelling requirement just as seriously as a hotel in Germany or Austria would. The effort involved is manageable; the risk of ignoring it is unnecessarily high.
What this means for social media content in practice
A simple framework helps in day-to-day work:
Clearly affected: fully AI-generated images (e.g. a room or product image created with Midjourney), AI-generated video clips, synthetic voices in ad clips, and heavily AI-altered photos that create a realistic impression.
Generally not affected: internal use of AI for text support, brainstorming, translations, or editorially reworked content where a human retains editorial responsibility and makes substantial changes.
The labelling itself doesn't need to be complicated. A clearly visible note that an image or video was AI-generated or AI-edited is sufficient in most cases. What matters is that it isn't hidden or buried in fine print.
Why this is an opportunity, not just an obligation
Interestingly, this regulation accelerates a trend that was already underway: the value of genuine, camera-produced content is rising. Once AI images have to carry a visible label, the distinction between "artificially generated" and "genuinely produced" becomes clearly visible to audiences for the first time, and more relevant to how a brand is perceived.
For businesses and hotels, this means: those who continue working with real photos and video don't need to worry about this requirement, and position themselves as authentic at the same time, in an environment that increasingly distinguishes between real and artificial content. That's not a coincidence; it's a shift worth positioning for early.
Conclusion
The AI labelling requirement starting in August 2026 isn't a reason to panic, but it is a good occasion to review your own content strategy: which content is currently created with AI tools, which with genuine camera production, and how transparently is that communicated to customers and guests? Clarifying these questions now, rather than improvising in August, saves stress, and strengthens credibility along the way.
This article does not constitute legal advice. For questions specific to your own situation, we recommend consulting a lawyer specialising in data protection and IT law.
