Legal
Last Updated On:
November 11, 2026
When GWI refers to AI in our products, we’re specifically talking about features that utilize large language models (LLMs) which encompass deep learning, machine learning, and natural language processing. It's important to understand that AI is integrated into limited, specific features within GWI’s platform, rather than being a pervasive element across all products or underlying technological infrastructure.
The GWI services use GWI’s proprietary consumer insights database only. No third party data or client data is used in any of the GWI services.
GWI Platform
Agent Spark (Platform)
GWI uses AI to deliver Agent Spark. This feature allows users to input their question into the search bar, generating interesting and relevant insights immediately. This is done without the users navigating or accessing other features of the platform directly. Agent Spark utilizes a large language model (OpenAI's GPT currently 5.4 and gpt-5-nano) via API to support the text generation of insights in response to the user’s question. It generates the outputs exclusively from GWI’s proprietary dataset (i.e. our database), ensuring that user inputs are never reused or surfaced in any output.
GWI Canvas
GWI also uses AI to deliver GWI Canvas. This feature allows users to generate a slide deck of insights within the platform. Users start with their goal (for example, media planning or campaign targeting) then define their audience, and GWI Canvas generates an insight-rich presentation packed with relevant charts, commentary, and key takeaways. It utilizes a large language model (currently OpenAI's GPT-5.0 and GPT 4 and 4.1 mini) to generate the content of the slides.
Agent Spark (MCP Connector)
Within all paid-for plans, GWI provides access to Agent Spark through a selection of pre-approved LLMs, including ChatGPT, Claude, Copilot Studio, and Gemini CLI. Users are able to access the pre-approved LLM and turn on GWI as a connector within the third party LLM using their GWI Platform login credentials. Then, much like in our platform, users can input their questions into the relevant LLM tool and generate insights immediately. The same large language model (set out above) is utilized in the process.
Integrations
Agent Spark (MCP Token)
Users can also connect Agent Spark through an unsupported third party LLM by utilising a GWI Spark API token or OAuth 2.0. The same GWI large language model (OpenAI's GPT set out above) is utilized in the process but the agent part is provided by the third party LLM.
Spark API
Spark API delivers GWI Spark insights via an API token.
The Spark API combines GWI data with natural language processing (currently by OpenAI’s GPT-5 via API, along with fine-tuned versions of GPT-4o and GPT-4o mini for specific tasks) to provide quick, actionable insights. It enables users to ask questions, such as "How do young parents discover brands?" and receive trusted, real world statistics on audience size, audience percentage, and index.
In terms of its relationship to other GWI products, Agent Spark, the Agent Spark MCP and Spark API deliver the same insights that are available from within the GWI Platform.
At the moment, users can choose not to use the AI features on the Platform. These features are entirely user-initiated, meaning they require active input or engagement from the user (for example, entering a prompt or generating a presentation) and are found in their own area of the platform. If a user chooses not to interact with these features, they remain inactive.
For full transparency, all AI-powered features are clearly labeled as such within the platform.
No, AI features are enabled by standard and this doesn’t have any impact on the platform’s performance.

All of GWI’s services - including databases that support our AI features - are hosted on Google Cloud Platform (GCP), specifically within the europe-west1 data center located in Brussels. For more information, see Google Cloud's data center locations.
This is currently the only hosting location we use. However, GWI retains the option to add additional regions in the future, for example to support disaster recovery, performance optimization, or other operational needs.
Access to logs and data within GCP is strictly controlled and protected by encryption both in transit and at rest. By default, Google-managed encryption keys are used, including customer-managed encryption keys (CMEK), as part of its internal security architecture where appropriate.. For further technical details, please refer to Google Cloud’s encryption documentation.
All outputs from Agent Spark (either: in the platform, Spark MCP and Spark API) follow the same GWI-side process below. For integrations, the GWI side flow remains the same but the difference is that the integrating LLM will render the final response (LLM Interpretation).
Agent Spark in the platform includes a thinking and source overview. In practice, this means the user can see a background actions summary together with structured metadata about the data used, including fields such as audience, topic, timeframe, location, and dataset, so the user can understand how the result was produced and what underlying GWI data it is drawing from.
Agent Spark has a documented evaluation framework covering end to end agent evaluation, audience-tool evaluation, insights-tool evaluation, and component-level evaluations.
GWI documents agent evaluation criteria that include query success, reasoning clarity, safety compliance and policy alignment, hallucination detection, chat coherence, and audience quality.
GWI also documents a dedicated grounding evaluation path that checks generated insights against underlying source data for numerical accuracy and text hallucinations.
GWI has also built in transparency by design in the documented flow:
The main documented checks currently available are:
These checks operate after or around generation as evaluation and validation layers over generated outputs.
The fact that GWI documents hallucination detection, numerical grounding checks, and source-attribution evaluation means the system is designed to detect and reduce risks such as unsupported claims, number mismatches, or incorrect source assignment. These are the kinds of issues the documented evaluation layers are intended to catch.
Given Agent Spark’s single source architecture and its grounding in real respondent data, the primary risk in this context would be the misrepresentation or misinterpretation of statistics, rather than the invention of entirely fabricated data. Nevertheless, users are always able to verify outputs, as all GWI insights are accompanied by a clear reference that can be easily accessed to cross check the underlying data and confirm consistency between the output and the source.
Both integration services proxy the insights generation to GWI's Spark Service Layer, following the same underlying flow as Agent Spark in the platform, intent extraction, database query, statistical calculation, and LLM interpretation of structured results.
LLM:
Spark MCP/Spark API:
LLM:
For clarification, the original prompt entered into the LLM such as ChatGPT (e.g., "Give me an analysis of how millennials and Gen Z prefer watching Premier League"), does not reach GWI's input database. GWI only collects the transformed prompts generated by the LLM (e.g. "search GWI for millennial watching Premier League").
As set out above, for platform services, GWI relies on ChatGPT for its infrastructure and ChatGPT confirms that API data is retained for up to 30 days.
For API services, your LLM of choice will set the retention period.
The AI model is hosted by OpenAI. GWI currently uses GPT-5 via API, along with fine-tuned versions of GPT-4o and GPT-4o mini.
GPT-4 model card is here
GPT 5 model card is here
Yes, GWI may fine tune responses received based on Agent Spark's inputs via the Platform. For example when a user searches for "insights about men in the US who own dogs" and GWI provides an insight, GWI may use the search to improve the accuracy and relevance of the insights generated. However, where Agent Spark is accessed via third-party tools, the prompts are not used to fine tune Agent Spark.
The Input database is used to power GWI Canvas, so this feature benefits from the fine tuning to identify the insights most useful to each audience.
GWI’s infrastructure and multi-layered safety mechanisms are designed to ensure that all AI-generated outputs are responsible, relevant, and aligned with the intended use of the platform.
When a user submits a prompt, it’s evaluated through a series of automated guardrails, including rule-based filters, OpenAI’s Moderation API, and LLM based relevance and safety classifiers. These checks help prevent misuse, ensure prompts are on topic, and harmful or inappropriate inputs are blocked.
Once cleared, GWI data (which is anonymized and aggregated) and the prompt are processed by the LLM. The model is used strictly to reformat inputs in order to surface insights from GWI data and ensure outputs are readable while respecting ethical and compliance boundaries.
Outputs are never generated from external or unvetted sources, and the system is continuously reviewed by our Product and Legal teams to maintain alignment with evolving ethical standards, regulatory requirements, and industry best practices.
Yes. AI features include built in safeguards to ensure responsible and stable use of the system. Each user is currently limited to one prompt per second and a maximum of 100 prompts per hour, helping to prevent misuse and ensure fair access across users. These thresholds may be adjusted as needed to maintain platform performance and reliability.
In addition, output length and complexity are also constrained by system level controls to ensure that results remain relevant, manageable, and aligned with the intended use of the platform. These measures form part of GWI’s broader commitment to responsible AI deployment.
GWI is implementing a multi-layered set of safeguards to prevent malicious or inappropriate use of free text prompting in AgentSpark, which is the only AI feature on the platform that accepts open ended inputs. These guardrails include:
These guardrails are being continually developed and strengthened to reflect best practices in AI safety. If a user input fails any of the GWI checks, it's not processed by the AI and the user is prompted to revise their input. Only inputs that pass all safety and relevance checks are handled by the AI agent.
No, because GWI doesn’t collect any personal data within its AI features (and no personal data or employee admin data is transferred to OpenAI).
Canvas is generated from GWI preset audiences and therefore there is no “input” that could include personal data.
Agent Spark has a prompt input box. These AI features aren’t intended to be used with personal data, and GWI makes it clear users should not input any personal data. All AI features “output” (based on either the preset audience generated by GWI or the prompt input audience) are anonymous and don’t include personal data, therefore falling outside the scope of GDPR.
GWI’s AI features don’t use sensitive or personal data. All insights generated by the AI features are based on GWI data, which is fully anonymized and aggregated. This means the outputs don’t contain any personal, identifiable, or sensitive information. Please refer to our main T&Cs for further information.
GWI doesn’t collect any personal data within its AI features (and no personal data or employee admin data is transferred to OpenAI). All services are processed within the European Union, specifically in the europe-west1 data center located in Brussels, via GCP, however, GWI reserves the right to add additional regions in the future. All data is encrypted both in transit and at rest, and GCP’s infrastructure meets strict security and compliance standards. For more information, see Google Cloud's data center locations and Google Cloud’s encryption documentation.
No. User inputs are never used as outputs for other users. Inputs are used to query GWI’s internal database and generate relevant insights for the user who submitted them, who may then opt to share these with others in their organization.
While inputs may be used internally to help fine-tune and improve the system’s ability to interpret prompts and improve database querying, all inputs are processed in an isolated environment, and outputs are generated exclusively from GWI’s proprietary data set, ensuring that user inputs are never reused or surfaced in any output.
Customers have full control over whether or not to use specific AI features as they are entirely user initiated, meaning they require active input or engagement and aren’t used unless the user chooses to. Customers can simply avoid these features if they prefer.
Inputs provided by users are used only to tailor the output and are never shared with other customers. While inputs may help improve the system’s performance internally (for example, by fine tuning how the model queries the database), this doesn’t involve exposing or reusing individual customer data in any way.
Our Legal and Product teams work together from development through to deployment to make sure GWI's AI features meet applicable legal and regulatory requirements and industry best practice (as these continue to develop).
GWI's AI features are made available to clients in the EU and are therefore within the scope of the EU AI Act. GWI is the provider of those features (as a “downstream provider”, because they are built on general-purpose AI models supplied by third parties, currently OpenAI). Based on the information available, those models are not classified as presenting "systemic risk". Any obligations that attach to the models themselves sit with the model provider.
None of GWI's AI features is a prohibited AI practice or a high-risk AI system under the Act:
GWI complies with its obligations under the EU AI Act as provider of its AI features. Our Main T&Cs support this: under clause 6.2, GWI commits to perform its obligations with reasonable skill and care and in accordance with applicable laws and regulations relevant to the Services.
Clients remain responsible for complying with the EU AI Act in their own capacity as deployers when they use the AI features. Our AI Acceptable Use Policy prohibits using GWI's AI features for high-risk use cases. Due to the nature of the services and AI features that GWI is providing and with this prohibition in place, we expect most client use cases to fall outside the high-risk category.
GWI monitors its obligations under the laws and regulations governing the AI landscape and remains committed to ongoing compliance. We are proactively strengthening our training, oversight, and governance frameworks to ensure ongoing alignment with current and future requirements.
GWI does not store the original prompts that clients submit to any third-party AI tools when using Agent Spark. For example, a prompt such as “What were Gen Z’s interests in energy drinks in 2025 for the UK?” is not received by GWI and is not stored in any GWI database.
Instead, the original prompts are processed and converted into structured queries (for example, “search GWI data for the following attributes: “Audience: Gen Z; Topic: Energy drinks; Timeframe: 2025; Location: UK”). These transformed, structured queries are then sent to GWI in order to retrieve relevant data. GWI stores these transformed queries in the GWI input database, but not the original free-text user input.
When GWI refers to "training on input data," this does not mean training a public or general purpose AI model. Instead, it refers to the internal use of user inputs to improve how our AI features interact with our database. This process helps us fine-tune the accuracy, relevance, and performance of our AI features when retrieving and presenting insights. Inputs are used solely to enhance the quality and efficiency of the service provided to GWI customers. This is done in a way that doesn’t expose inputs to other clients.
The client retains all intellectual property and other rights in the AI inputs. GWI owns all intellectual property and other rights in the AI outputs (as these are insights based on GWI’s proprietary data). As set out in clause 2 of our Main T&Cs, clients have a right to use the feature outputs under the licence granted. The AI features are used to surface GWI data in a way that’s more intuitive for users, but it remains GWI’s intellectual property, ensuring GWI can continue to offer access to GWI data to all clients via its various products.
All AI outputs are generated from insights collected and owned by GWI, meaning they don’t infringe third party copyrights. However, when using GWI data externally, it must be properly cited in accordance with our referencing guidelines.
As standard, AI outputs are subject to the same terms as all GWI data under our main T&Cs. Unless otherwise expressly set out in your order form, you can only use GWI data for your own internal business purposes, and for limited external business purposes. External use is allowed where it involves referencing or quoting GWI data in the context of your usual business activities - for example, media planning, campaign reporting, creative development, marketing, or strategy work for clients. If you are interested in exploring further external use cases, please speak to your sales representative as alternative options are available under GWI’s data distribution T&Cs.
It’s the customer’s responsibility to determine whether any disclosures are required when using AI outputs in their work, and to communicate them as appropriate within their organization. As with all services offered by GWI, customers are expected to comply with all applicable laws and regulations, including any obligations related to transparency or user notification. Please refer to our main T&Cs for further guidance.
Clients may use GWI Data in any AI tools for internal business purposes, provided they:
If you’re considering an external or alternative use case, please contact your sales representative so we can assist you.