- European Commission has begun enforcing AI transparency rules from the 2024 AI Act, imposing hefty fines on rule-breakers.
- Under the new transparency requirements, companies must clearly label AI-generated or manipulated content.
- The EU has introduced specific icons such as "AI", "AI Generated", and "AI Modified" for labeling.
- Labeling applies to various content types, including images, audio, and video that resemble real entities or events.
- Companies must inform users when they are interacting with an AI system instead of a real person.
- Fines for non-compliance can reach up to €15 million or 3% of annual turnover.
- The European Commission has gained new powers to investigate and impose fines on AI companies.
- OpenAI and Anthropic are now under scrutiny following recent cyber attacks involving their models.
- OpenAI has confirmed it is in contact with the EU AI Office regarding compliance.
- The EU AI Office has supervisory and enforcement powers over general-purpose AI models as part of the 2024 AI Act.
- Henna Virkkunen from the European Commission highlighted the risks associated with improperly designed AI.
- Experts warn that GPAI liability extends beyond substantive breaches to include misleading information and non-compliance with information requests.
The European Commission has initiated enforcement of the AI transparency rules established by the 2024 AI Act, targeting companies like OpenAI and Anthropic.17
Under these new regulations, companies must clearly label AI-generated or manipulated content, including images, audio, and video.2
The EU has introduced specific icons such as "AI", "AI Generated", and "AI Modified" to facilitate compliance.3
Failure to adhere to these rules could result in fines of up to €15 million or 3% of annual turnover.6
EU institutions can face fines up to €750,000, with smaller companies subject to proportional penalties.
The Commission's new powers allow it to investigate and impose penalties on AI developers, ensuring that models are evaluated before public release in the EU.
Henna Virkkunen, executive vice-president for tech sovereignty, security and democracy at the European Commission, emphasized the risks associated with poorly designed AI.11
OpenAI has confirmed its cooperation with the EU AI Office, while experts warn that non-EU companies must also comply with these regulations.910
Elisabetta Righini, a partner at Sidley Austin, noted that liability extends beyond substantive breaches to include misleading information or obstructing evaluations.
As the EU implements these rules, companies like Google express commitment to compliance and advancing AI innovation in Europe.
“Companies that fail to label AI-generated content face fines of up to €15 million or 3% of annual turnover, while EU institutions face penalties up to €750,000. The rules also require clear labeling of deepfakes and chatbot disclosures, and non-EU providers must appoint an EU-based authorized representative.”
CNBC
