Generative AI Policy

Introduction
This AskNicely Generative AI Policy (“Gen AI Policy”) outlines AskNicely’s approach to using generative artificial intelligence (AI) tools and systems. The Gen AI Policy applies to AI-enabled functionality within the proprietary AskNicely Customer Experience Platform (“AskNicely Platform”) that is made available to customers. This includes how we design, test, deploy, monitor, and govern AI features, and how we handle customer data when AI capabilities are used. In addition, the Gen AI Policy applies to the use of AI in other operations across our organisation. When referring to generative AI, we mean tools that create new text, images, audio or video, based on prompts. We recognise both the transformative potential of generative AI and our responsibility to use these technologies ethically, safely, and in accordance with government guidance.

Headquartered in Portland, Oregon, in the United States, AskNicely operates globally through its affiliates. We operate the AskNicely platform offered as both a web-based service and mobile app, the website located at , and associated services (collectively referred to as “Services”).

Purpose
The purpose of this policy is to: 

• Establish clear guidelines for the responsible use of generative AI at AskNicely, including Ask Nicely Holdings, Inc., and our affiliates, Ask Nicely LLC, Ask Nicely, and Ask Nicely B.V. (collectively “AskNicely”) and its employees and AskNicely Platform users
• Ensure our AI practices align with government guidelines 
• Provide transparency on our use of AI technologies to our customers
• Mitigate potential risks associated with generative AI
• Uphold AskNicely’s core values and company policies including AskNicely’s Privacy PolicyPrivacy Policy

Customer Assurance
AskNicely is committed to operate the AskNicely Platform with assurance that customer data will be handled and processed within the following guidelines:

• Customer data is not used to train AI models
• There is no cross-customer data mixing and this is per the design of the AskNicely Platform and does not require any customized settings
• Human review is deployed for oversight of AI usage and generated content
• Incident response involves a clear line of communication and compliance with an outlined response process

Scope
This Gen AI Policy applies to (1) all AskNicely employees and contractors using generative AI in their work, and (2) all AI-enabled features offered within the AskNicely Platform. It covers all generative AI applications, including but not limited to:

• Text generation tools (e.g., ChatGPT, Claude)
• Image generation tools (e.g., DALL-E, Midjourney)
• Audio/video generation tools (e.g., Notebook LLM)
• Use of third-party AI tools by employees (e.g., writing, analysis, prototyping)
• Use of AI to assist with the AskNicely Platform development (e.g., code assistance, code review, testing, debugging, and documentation)
• Design and operation of customer-facing AI functionality within the AskNicely platform (e.g., AI-assisted insights, summaries, classifications, recommendations, or similar)
• Any other generative AI technologies, including future technologies of a similar nature to make sure the policy applies as technology evolves

Guiding Principles
Human-centered Approach
AskNicely takes a human-centred approach to its AI use. This is achieved by designing AI use to prioritize human value and human rights, provide human oversight across the AI cycle, and use AI as a tool to augment human capabilities rather than replace human judgment. In addition, we have implemented human review on decisions involving AI outputs that significantly impact individuals. Technology capabilities are assessed to prioritize our customers and other stakeholders. AskNicely regularly reviews evolving technology capabilities before integrating them into its business use and/or AskNicely Platform.

Transparency
To ensure we stay aligned with our dedication to ethical AI tool integrations AskNicley commits to be transparent within our company and to our customers.

AskNicely communications and documents‍

• Disclose AI assistance where it is material to the deliverable, methodology, or customer decision-making.
• Ensure a human reviewer is accountable for externally shared content.
• Maintaining transparency with customers about our use of AI tools
• Explaining AI-assisted decisions in understandable terms
• Documenting AI processes as needed
• Labelling AI generated images as such according to the IPTC Photo Metadata Standard

AskNicely Platform‍

• Clearly labelling AI-assisted functionality where appropriate
• Providing feature-level explanations of what the AI does, its intended use, and known limitations
• Offering customers a way to raise questions or concerns about AI-enabled functionality

Privacy and Data Collection
AskNicely designs the features in the AskNicely Platform to comply with AskNicely’s Privacy Policy and applicable data protection regulations. AskNicely does not provide customer data to public consumer-grade AI services in a way that permits those services to retain, reuse, or train on that data. Nor is customer data used to train general-purpose AI models. Any AI-assisted content that is externally shared undergoes human review. AskNicely invests in paid subscriptions for tools that process personal data and implements measures such as data protection agreements governing data processing.

Accountability and Governance
AskNicely commits to:

• Designating clear roles and responsibilities for AI oversight within our organisation
• Reviewing our AI policy as part of our regular policy review, or sooner if needed to reflect evolving best practices
• Maintaining records of AskNicely’s significant AI use cases and their impacts
• AI-enabled product features must undergo an internal review and approval process prior to release (including privacy, security, and quality considerations)
• Material changes to AI-enabled functionality (e.g., model/provider changes, new data inputs, or new use cases) must be assessed for risk, and documented and tested appropriately
• Adhere to all copyright laws and recognizes that AI may generate content that is not wholly unique to the prompt, and human oversight will be deployed to ensure that AI content is not in violation of copyright

Prohibited Uses
AskNicely employees must not:

• Input customer data into public consumer-grade AI services that use such data to train models
• Use AI outputs as a substitute for required approvals, legal review, or security review
• Represent AI-generated content as verified fact without validation
• Use AI in ways that circumvent customer contractual requirements or privacy obligations
• Use AI to generate explicit, violent, or other such content outside the bounds of appropriate workplace conduct
• AskNicely users are responsible for the AI content they generate and utilize on the platform and are prohibited from creating and utilizing explicit, graphic, and harmful content

AI in the AskNicely Platform (customer facing AI)
‍AskNicely may provide AI-enabled features as part of the AskNicely Platform to help customers derive value from their CX data (e.g., generating summaries, surfacing themes, assisting analysis, creating dynamic surveys or similar AI-supported capabilities). We take a risk-based approach to ensure these features are reliable, safe, and appropriately governed.AI-enabled features are designed for specific, documented use cases. Where outputs may influence customer decisions, we provide guidance and design controls to support appropriate human review.

Data Boundries and Tenant SeperationAskNicely commits to the following safeguards for customer data:

• The commitments in Section 5.3 (Privacy and data protection) apply to all AI-enabled platform features.
• Customer data is logically separated by tenant and access-controlled. Where shared AI models are used, controls are in place to prevent cross-customer data access and mixing.
• Access to customer data for AI processing is limited to what is necessary to deliver the specific feature and is governed by least-privilege access controls.
• AskNicely AI Testing Framework plus monitoring for every AI feature offered in the AskNicely Platform.
• Feature-level documentation of intended use in the AskNicely Platform plus limitations

Testing, Quality Assurance, and Limitations of AI Outputs in the AskNicely Platform
Generative AI and probabilistic models can produce outputs that are incomplete, incorrect, or unexpected. In addition, we are aware that AI may generate content that is nonfactual, representative of harmful biases, or of an explicit nature. AskNicely cannot validate every possible output of its platform for every possible input. However, we use a structured testing and assurance approach before and after release of AI-enabled functionality, including:

• Prompting and system instructions are designed with guardrails appropriate to the feature, including input constraints and output validation where applicable
• Defined use cases and acceptance criteria for each AI feature
• AskNicely’s AI Testing Framework includes feature-level test plans, representative scenario evaluation, regression testing, and production monitoring appropriate to the risk level of each AI feature.
• Evaluation for accuracy and safety using representative scenarios and edge cases relevant to the feature’s purpose
• Monitoring and incident response for AI features in production, including investigation and remediation when issues are detected
• Regression testing when prompts, models, or providers change
• Rollback / feature flagging where appropriate
• Monitoring designed to detect degraded performance or unexpected behaviour over time (including model drift where applicable)

Where appropriate, AskNicely includes product guidance reminding users that AI-assisted outputs should be reviewed and validated in context, particularly for high-impact decisions.

Training and Awareness
All team members at AskNicely receive training on this Gen AI Policy and responsible AI use. AskNicely distributes regular updates on AI best practices to team members, including scenario-based learning. AskNicely provides team members with multiple avenues of reporting to provide feedback, raise concerns, and contribute to the optimization, effectiveness, and enhancement of AI use in the AskNicely Platform.

Incident Response
An AI event is an event that includes:

• Breaches data privacy or security requirements as defined in AskNicely’s Privacy Policy and Security Policy
• Produces materially inaccurate, biased, explicit in nature, containing graphic violence, harmful, discriminatory, or culturally unsafe outputs
• Contravenes customer contractual obligations or regulatory requirements

In the event of an AI Incident, the AskNicely’s incident management procedure will be followed, including, reporting the AI Incident to AskNicely senior management, documenting and verifying the AI Incident and its impact. Depending on the assessment, AskNicely will take remediation steps and review and updates the Gen AI Policy as needed. External reporting will comply with contractual obligations and regulatory requirements.

Policy Review
‍The current version of the Gen AI Policy will be reflected in the “Last updated” date at the top of the document. The current version of the Gen AI Policy will govern our approach and use of AI and will always be published at https://www.asknicely.com/ai-policy. The Gen AI Policy will be reviewed annually or more frequently as AI technology and guidelines evolve.