Glossary

EU AI Act

Regulation (EU) 2024/1689, the first comprehensive AI law. It bans some uses, places heavy obligations on high-risk systems, adds transparency duties for chatbots and generated content, and regulates general-purpose models, in stages from 2025 to 2027.

General definition

The EU AI Act is Regulation (EU) 2024/1689, published in the Official Journal on 12 July 2024 and in force since 1 August 2024. It applies to providers who place AI systems on the EU market, to deployers who use them in the EU, and to providers outside the EU whose system’s output is used in the EU, wherever the company is based. Like the GDPR, it reaches well beyond European companies. The Act is risk-based: obligations attach to how a system is used, not to the technology as such.

  • Unacceptable risk (Article 5): prohibited practices such as harmful manipulative techniques, social scoring that leads to unjustified detrimental treatment, untargeted scraping of facial images, and emotion recognition in workplaces and schools
  • High risk (Annex I and Annex III): AI in regulated products such as medical devices, and listed uses such as credit scoring, life and health insurance risk pricing, recruitment, education, critical infrastructure, emergency call and patient triage, and law enforcement; these need risk management, data governance, technical documentation, logging, human oversight, accuracy and robustness, a conformity assessment and registration
  • Transparency risk (Article 50): systems that interact with people must make clear they are AI unless it is obvious; deepfakes and AI-generated text on matters of public interest must be labelled
  • Minimal risk: everything else, with no new obligations beyond existing law

General-purpose AI models (GPAI), the foundation models behind most chatbots, have their own chapter. Providers must keep technical documentation, give downstream integrators the information they need, publish a summary of training content and respect EU copyright law. Models with systemic risk, currently identified by a training-compute threshold, must also run evaluations, report serious incidents and secure the model. A voluntary Code of Practice helps providers demonstrate compliance.

The timeline is staged. Prohibitions and the AI literacy duty in Article 4 applied from 2 February 2025. GPAI obligations, the governance structure and the penalty regime applied from 2 August 2025. Most remaining rules, including Annex III high-risk systems and the Article 50 transparency duties, apply from 2 August 2026, with high-risk AI embedded in Annex I regulated products following on 2 August 2027. The European Commission proposed in late 2025 to push some high-risk deadlines back, so check the current status before planning around a date. Fines reach EUR 35 million or 7% of worldwide annual turnover for prohibited practices and EUR 15 million or 3% for most other breaches.

In the Ethora ecosystem

For a company deploying an AI chat agent, the Act usually lands in the transparency tier: an AI persona that talks to customers must be identifiable as AI, and staff who operate it need basic AI literacy. If the agent is used for a listed high-risk purpose, for example deciding on credit or triaging patients, the deployer duties in Article 26 apply: use the system as instructed, assign human oversight, keep the logs it generates for at least six months, and in some cases run a fundamental rights impact assessment alongside the DPIA the GDPR already requires.

Ethora gives a deployer the controls those duties assume. Agents are configured per room with a visible name and system prompt, so disclosure is a matter of configuration rather than retrofitting. Conversations are archived server-side with configurable retention, and the compliance audit trail with immutable export covers the logging duty. Per-agent model choice means you can run an open-weight model on your own infrastructure when a GPAI provider’s documentation or data terms do not fit. Ethora is a platform provider, not your conformity assessor: the classification of your use case and the compliance file behind it stay with you, as part of your AI governance.

Get started

Chat and AI built for regulated industries

Ethora gives you audit trails, retention controls, a Trust & Safety layer and dedicated deployments in your own cloud. Talk to our team.

Start Free
Free tier available Enterprise SLA No vendor lock-in