The European Union's AI Act has begun its initial phase of enforcement, introducing new requirements for transparency in AI systems, particularly those interacting with consumers. As of August 2, providers of chatbots, deepfakes, and other AI-generated content are now subject to rules mandating clear disclosure of their artificial nature. This includes labeling AI-generated audio, images, video, and text, and requiring chatbots to identify themselves as such.
These transparency obligations extend to general-purpose AI (GPAI) models, which encompass large language models and other foundational AI systems. The newly established AI Office in Brussels, in collaboration with national regulatory bodies, is responsible for enforcing these provisions. Content published on or after August 2 must adhere to these labeling requirements, even if it was generated prior to that date.
The Act also formally prohibits certain AI uses deemed to pose an "unacceptable risk." These include AI systems designed to generate non-consensual sexually explicit or intimate content, such as "nudifier" applications, and child sexual abuse material. While these specific prohibitions will not be enforced until December 2, 2026, giving companies time to comply, the immediate transparency rules carry significant penalties. Non-compliance can result in fines up to 15 million Euros (or 3% of worldwide annual turnover, whichever is greater).
To facilitate enforcement, the AI Office has launched several tools. Individuals and organizations can utilize a complaint tool to report alleged infringements by AI system providers or deployers. A whistleblower tool offers an anonymous channel for professionals within AI companies to flag potential violations that could impact fundamental rights, health, or public trust. Additionally, a downstream complaints channel allows firms that build upon GPAI models to report suspected breaches by the underlying model providers, creating a formal mechanism for addressing systemic issues like unsafe model behavior or ignored red-team findings.
The intent behind these transparency measures is to mitigate the deceptive potential of AI by requiring clear identification, rather than outright banning the technology. While the Act acknowledges that malicious actors may still use unlabeled tools outside the EU, it aims to raise the bar for legitimate services and ensure mainstream platforms address the risks associated with deceptive AI features. The implementation of the AI Act marks a shift towards a regulated environment for AI within the EU, moving from a largely unregulated landscape to one with basic operational rules.
