At Howspace, we believe that artificial intelligence should empower, not replace human decision-making. That’s why we’ve integrated AI into our digital facilitation platform in a way that’s responsible, transparent, and fully aligned with the European Union’s Artificial Intelligence Act (EU AI Act, Regulation (EU) 2024/1689).
We deliver our AI capabilities through an enterprise-grade large language model service, hosted in the EU, with strong contractual safeguards on how customer content is handled.
A risk-based, responsible approach to AI
The EU AI Act introduces a risk-based regulatory framework to ensure that AI systems in the EU are safe, lawful, and respect fundamental rights.
Howspace does not use AI for high-risk purposes as defined in the EU AI Act (e.g., employment, credit scoring, biometric surveillance, or law enforcement).
Instead, our AI features, such as summarization and prompting, are designed to assist users in collaboration and sense-making. These are classified as low-risk AI use cases.
As a result, Howspace is subject to transparency and general use obligations under the Act, not the stricter requirements reserved for high-risk systems.
Howspace’s role under the EU AI Act
The EU AI Act distinguishes between organisations that develop and place AI systems on the market (“providers”) and organisations that use AI systems under their own authority (“deployers”). When AI capabilities are built by one organisation on top of a general-purpose AI model offered by another, both roles can apply at the same time.
That’s the case for Howspace:
- Howspace is a provider of the AI capabilities offered within the Howspace platform. The Summary, Word Cloud, Mindmap, AI+, transcript and other AI features are downstream AI systems built into the platform and placed on the market under the Howspace brand.
- Howspace is concurrently a deployer of the underlying general-purpose AI model that powers those capabilities. We obtain that model as a service from an AI sub-processor. We do not develop, train or fine-tune the underlying model.
Built on an EU-hosted enterprise LLM service
All AI functionality in Howspace is powered by an enterprise large language model service hosted in the EU, with a separate regional endpoint for our Canadian customers. The underlying provider implements layered compliance commitments that align with the EU AI Act, including:
- Proactive implementation of the AI Act’s requirements for general-purpose AI (GPAI) models
- A Responsible AI standard and restricted-use policies that mirror the AI Act’s prohibited practices provisions
- Updated contracts and Codes of Conduct that prohibit unlawful uses (e.g., emotion recognition in the workplace, social scoring, or biometric surveillance)
- Transparency Notes, documentation, and tooling that support downstream compliance
Howspace benefits from these safeguards as a downstream user and builds additional safeguards on top. The specific AI sub-processor in use at any given time is identified in our Data Processing Agreement, which forms part of every customer contract.
We do not train AI models on your data
Customer content is not used to train any Howspace, supplier-controlled or third-party AI model. This commitment binds Howspace and the AI sub-processor that powers our capabilities. Prompts and completions sent for processing are not retained by the model after the response is generated.
Transparency and user awareness
In line with Article 50 of the EU AI Act, Howspace ensures that:
- Users are informed when they are interacting with AI features (e.g., AI-generated summaries or clustering).
- AI-generated content is distinguishable and appropriately explained in context.
- Documentation is available to help customers understand the purpose, functionality, and limitations of Howspace’s AI features.
The current list of AI capabilities offered within the platform is published on our Trust Center. The list is exhaustive at the platform level; which capabilities are actually available to a given customer depends on the customer’s Howspace subscription and on the sub-processors the customer has not objected to under the Data Processing Agreement.
What we don’t do
Howspace does not use AI for any of the following prohibited or high-risk purposes:
- Emotion recognition or biometric categorization
- Real-time or remote biometric identification
- Criminality prediction or social scoring
- Employment or credit-related decision-making
These practices are either prohibited outright or subject to strict regulation under the AI Act. We have explicitly excluded them from our product roadmap.
Contracts, controls, and commitment
We provide contractual clarity and safeguards to our enterprise customers:
- Our use of Azure OpenAI is governed by Microsoft’s updated terms and Code of Conduct.
- We are committed to ongoing monitoring of legal developments and updates to the AI Act.
- We provide customers with clear data handling and security documentation, aligned with GDPR and ISO/IEC 27001 standards.
Responsible AI is a shared journey
Howspace embraces the EU AI Act as a forward-looking framework that supports innovation while protecting people. Together with Microsoft, we are building AI-powered collaboration tools that are safe, human-centric, and fully aligned with European values.
For more information, visit: