Skip to main content
The EU AI Act (Regulation (EU) 2024/1689, known in German as the KI-Verordnung or KI-VO) sets obligations for providers and deployers of AI systems, including transparency duties for AI systems that interact with people.
For questions about your obligations under the AI Act, email legal@botbrains.io. The binding terms are in § 11 of the Terms of Service, this page summarizes them.

Provider and deployer roles

botBrains develops and provides the AI agents, so botBrains acts as the provider of these AI systems (Art. 3(3)). botBrains doesn’t provide general-purpose AI models (Art. 3(63)). When you run a botBrains agent, you act as the deployer (Art. 3(4)): you determine the deployment context, the data, and the content.

Transparency

botBrains meets the transparency obligation under Art. 50(1) by informing users that they’re interacting with an AI system wherever botBrains controls the user-facing surface, such as the chat widget. When you integrate the agent into your own systems or channels, such as email or ticketing, you take on the Art. 50 transparency obligations, including labeling AI-generated or AI-assisted content. botBrains provides the technical documentation to implement these notices for Slack, Zendesk, and Salesforce Service Cloud.

Prohibited and high-risk use

botBrains builds its agents for customer support and Q&A and doesn’t design or operate them for high-risk use cases. You must not deploy them for the practices the EU AI Act prohibits or classifies as high-risk. Prohibited practices (Art. 5). You must not use the agents to, for example:
  • manipulate or deceive users to distort their decisions to their detriment, or exploit vulnerabilities related to age, disability, or socio-economic situation;
  • score people based on their social behavior (social scoring);
  • categorize people by biometric data to infer sensitive traits such as race, religion, political views, or sexual orientation;
  • predict the risk of someone committing a crime based on profiling, or build facial-recognition databases by scraping images.
High-risk use cases (Art. 6 and Annex III). You must not deploy the agents as part of, for example:
  • recruitment or HR decisions, such as screening applications or deciding on promotion or termination;
  • monitoring or evaluating the performance or behavior of employees, see Grading employees;
  • creditworthiness or credit scoring, or insurance risk assessment and pricing;
  • deciding eligibility for essential public benefits or services, or emergency-call triage and dispatch;
  • admission to education or the evaluation of exams;
  • safety components of critical infrastructure, such as water, gas, electricity, or road traffic;
  • law enforcement, migration, asylum and border control, or the administration of justice and elections.
You assess whether your intended use could fall under Art. 6; any high-risk use requires a prior written agreement with botBrains, and you carry the resulting obligations, in particular under Art. 26. botBrains may suspend or terminate the service if you breach these limits.

Insights and ServiceLens

Insights and ServiceLens analyze your customer conversations, including the ones only your team handled. That becomes employee grading as soon as AI judges how well an identifiable person on your team works. It happens, for example, when an AI attribute, a monitor with flag criteria, or a scorecard built on AI attributes rates the answers in conversations your team handled, or when you compare such results by employee. Employee grading by AI is high-risk under the EU AI Act, and botBrains doesn’t allow it. If you plan to analyze your team’s conversations, read Grading employees first.

Sentiment and emotion recognition

Conversation attributes and monitors read the transcript, so a sentiment attribute analyzes words, not biometric signals. Inferring employees’ emotions from their voice or face is a prohibited practice in the workplace under Art. 5(1)(f), apart from medical or safety reasons. A customer’s frustration or a failed conversation doesn’t, on its own, show poor performance by an employee.