Responsible AI in Practice

What Is Responsible AI?

Responsible AI is the practice of designing, developing, and deploying artificial intelligence with the intention to empower users and society. It emphasizes:

• Trust

• Safety & Security

• Accountability

• Reliability & Robustness

• Explainability & Transparency

These pillars are detailed out in our Unique AI Governance Framework (AI Governance Framework). When applied thoughtfully, AI can be both ethical and beneficial — enhancing outcomes while respecting users and communities.

What Does It Mean for Me? Intended use cases

Use Unique AI responsibly and lawfully.

The Service is designed exclusively for use in the financial services sector.

You must not use the Service for purposes outside the intended financial service use cases. Prohibited uses include, but are not limited to:

• Credit scoring

• Biometric identification

• health‑related consultation, diagnosis, or treatment,

• provision of therapeutic or medical advice,

• use in safety‑critical or life‑critical decision making,

• or any use cases that contravene applicable laws, regulations, or ethical AI standards.

Unique AI adheres to Responsible AI principles that govern the development and use of its platform. By using the Service, you agree to comply with these practices as set out in our Responsible AI Policy (Unique AI Policy). Any use of the Service in violation of these practices is strictly prohibited.

Subprocessors and ‑Third Party‑ Providers: Unique AI engages subprocessors and Third-Party large language model (LLM) providers in order to deliver the Service. You agree to comply with any responsible AI practices and usage restrictions that apply to such ‑subprocessors‑ and providers, including respective Data Processing Agreements in addition to those of Unique AI.

The Service is for informational purposes only and does not constitute financial, investment, tax, or legal advice. Consult qualified and appropriate professionals before making related decisions.

If You See Something, Say Something

If you witness or suspect misuse — such as offensive content, hacking attempts, or other harmful behavior — report it immediately. We take all reports seriously to preserve the integrity and safety of our service.

Creating & Sharing Content

You can create and share content through the chat as long as it follows these principles:

✅ Acceptable

• Respectful, inclusive, and lawful content

• Original or properly attributed materials

• Within the data sharing principles of your organization

🚫 Unacceptable

• Offensive or discriminatory language

• Defamatory or rights-infringing content

• Anything promoting hate, harm, or violence

When in doubt, err on the side of caution.

What Counts as Malicious or Harmful Use?

Malicious content includes anything designed to harm the platform, users, or data, such as:

• Malware or executable code

• Prompts meant to extract sensitive information

• Attempts to access PII, CID, or other private infrastructure

Consequences of Malicious Activity

• Immediate suspension or termination of access

• Potential legal action

• Internal review of the breach

If you accidentally submit something harmful, report it immediately so we can mitigate any impact.

Data Privacy & Prohibited Inputs

Unless expressly allowed by you organizations CISO, Compliance and IT teams, never transmit the following to the chat:

• Personally Identifiable Information (PII)

• Client-identifying data (CID)

• Any personal or confidential data you are not authorized to share

• Information restricted by law, contract, or third-party rights

If You Accidentally Share Sensitive Data

Notify us right away. We’ll take appropriate measures — but remember, you are responsible for the data you input.

Security Testing & Technical Boundaries

Do not perform:

• Load testing

• Penetration testing

• Any testing that could degrade performance

Unless you have explicit written approval. Details are usually specified in the contract.

Human Oversight & Content Accuracy

• Always review, adjust, or modify AI-generated outputs before use.

• Do not rely on AI to be fully accurate or legally reliable.

• Use human judgment, especially for public, sensitive, or compliance-related content.

You — the user — are ultimately accountable for all actions taken based on the AI’s output.

Questions or Reports?

For incident reports or guidance, please contact your designated Unique AI administrator or security contact: aigovernance@unique.ai.