At Spacewell Energy, we are committed to the responsible development and use of Artificial Intelligence (AI). This article outlines how our platform and services comply with Regulation (EU) 2024/1689, commonly known as the AI Act, which establishes the rules for using AI in the European Union.
Overview of the AI Act
The AI Act regulates the development and deployment of AI models and systems based on their level of risk. Based on a comprehensive diagnostic analysis, we have determined that the AI systems currently used or provided by Spacewell Energy do not constitute a high risk to users or affect fundamental rights.
Classification of Our AI Systems
Spacewell Energy utilises several AI systems to enhance energy management tasks. These are categorised as Specific-Purpose AI because they are designed for clearly defined technical functions in the energy field.
- Proprietary Systems: We have developed four internal tools: Optimise (anomaly detection), Detect (smart building comparison), Forecast (energy consumption prediction), and ABC (baseline creation).
- Third-Party Systems: We integrate specialised tools for support and internal efficiency, including FIN AI Chat (Intercom, Inc.) and Zendesk AI (Zendesk, Inc.).
Our analysis concludes that none of these systems falls under the "Prohibited", "High Risk" or "Systemic Risk" categories regulated by the AI Act.*
*In the case of third-party systems, according to their guidelines.
Our Roles Under the AI Act
Depending on the system, Spacewell Energy assumes different regulatory roles:
- Provider: We act as the provider for our proprietary AI systems (Optimise, Detect, Forecast and ABC), ensuring they meet technical standards for reliability and performance.
- Deployer: We act as a deployer when using third-party AI systems like Intercom and Zendesk for professional purposes.
Transparency and User Interaction
The AI Act emphasises the importance of transparency so users know they are interacting with AI.
- No Synthetic Content: Our AI systems generate analytical results (mathematical calculations, statistical graphs and reports) rather than "synthetic content" (text, images or audio) that could be confused with reality.
- Clear Disclosure: While the risk of confusion with a natural person is minimal, we maintain best practices by providing disclaimers in our interfaces to inform users when responses are automated.
Data Governance and Training
We prioritise the security and lawfulness of the information used to train our models, ensuring that all processes align with rigorous legal standards.
- Regulated Data Usage: The use of customer data for model training or fine-tuning is strictly governed by the specific contractual conditions and agreements established between Spacewell Energy and our clients.
- Focus on Non-Personal Data: Our internal systems primarily utilise non-personal, technical data, such as energy consumption patterns and meteorological information, which does not fall under personal data protection regulations.
- Lawful Sourcing: All third-party information integrated into our AI systems is obtained through professional licenses, ensuring that our data sources are fully authorised and compliant with intellectual property requirements.
- Confidentiality Safeguards: We implement professional-tier subscriptions for third-party AI tools to guarantee that internal data is protected and never reused for the providers' own model retraining.
Our Ongoing Commitment
Spacewell Energy follows industry best practices to ensure our AI solutions remain robust and secure:
- Regular Monitoring: We supervise the functioning of our algorithms and perform regular performance validation.
- Human Oversight: We maintain human oversight mechanisms to record incidents and ensure systems function as intended.
- Cybersecurity: We apply rigorous technical standards, including ISO 27001, to protect our systems and your data.
For more detailed information regarding our data protection policies, please refer to our Privacy Policy or contact our support team.