AI and Technology Disclosure Effective Date: October 31, 2026 1. You Are Interacting with AI U Navigate uses artificial intelligence throughout the Platform. This disclosure explains what AI does on the Platform, what it cannot do, and what you should know before relying on any AI-generated content. Lexi is an AI assistant, not a human. When you interact with Lexi in the assessment flow or in chat, you are communicating with an AI language model. Lexi is designed to sound natural, helpful, and conversational. This is intentional and does not mean Lexi is a person. You are always talking to software. We disclose this clearly and without ambiguity because we believe you have the right to know when you are interacting with AI, especially when discussing sensitive personal circumstances. 2. How AI Is Used on the Platform Assessment and Intake The conversational assessment is guided by AI. Lexi generates contextual question language, acknowledgment responses, and dynamic follow-up questions based on what you share. These responses are generated in real time and personalized to your situation. Support Plan Generation Your support plan is generated by a large language model (AI) using your assessment responses, demographic information, and geographic location. The plan includes barriers identified, recommended next steps, localized resources, and stability indicators. AI-generated plans are produced from patterns in training data and the specific inputs you provide. They are not the product of professional assessment by a licensed social worker, counselor, healthcare provider, or attorney. Resource Recommendations Resources shown on the Platform come from a catalog that is curated and verified by people, not selected or written by AI. When the Platform suggests you may qualify for a specific benefit, that suggestion is produced by fixed eligibility rules applied to information you deliberately provided — it is not an AI guess, and the reason for the match is shown with it. Resource details such as phone numbers, addresses, and availability are re-verified on a schedule but can still become outdated; always confirm directly with a program before relying on it. Chat Responses Responses from Lexi in the chat interface are AI-generated. Lexi can use context you have shared on the Platform — your profile basics, your tasks, documents you reference, and your assessment results — to personalize responses. Lexi also uses U Navigate's program information, and for programs matched to you, the reasons they matched. When web search is enabled for a message, Lexi can retrieve current information from the internet; otherwise, her knowledge of the wider world comes from training data and may be out of date. Lexi cannot independently verify a program's current availability, waitlist, or eligibility rules; always confirm important details directly. 3. What AI Cannot Do You should not use the Platform as a substitute for professional advice. Specifically: • Lexi is not a licensed social worker, counselor, therapist, psychiatrist, or healthcare provider • AI-generated content is not medical advice, mental health treatment, or crisis intervention • AI-generated content is not legal advice and does not establish an attorney-client relationship • AI-generated content is not financial advice • Lexi cannot access real-time information about program availability, waitlists, or eligibility • Lexi cannot guarantee the accuracy of resource phone numbers, addresses, or hours • Lexi cannot predict outcomes or guarantee that following a support plan will produce any specific result • Lexi is not able to call emergency services on your behalf 4. Emergencies and Crisis Situations IF YOU ARE IN IMMEDIATE DANGER OR EXPERIENCING A MEDICAL EMERGENCY, CALL 911 IMMEDIATELY. DO NOT RELY ON THIS PLATFORM FOR EMERGENCY RESPONSE. If you are experiencing a mental health crisis, suicidal thoughts, or thoughts of harming yourself or others: • Call or text 988 (Suicide and Crisis Lifeline) — available 24 hours a day, 7 days a week • Text HOME to 741741 (Crisis Text Line) • Call 911 or go to your nearest emergency room If you are in an unsafe domestic situation: • National Domestic Violence Hotline: 1-800-799-7233 (call or text, 24/7) • Text START to 88788 The Platform may provide these resources within a support plan, but providing them here is not a substitute for contacting emergency services directly. How Lexi responds when things are heavy If you tell Lexi something that sounds like you may be thinking about hurting yourself, or that someone is hurting you, Lexi is built to respond in a specific way: she stays in the conversation, she makes sure you know you can call or text 988 any time and text HOME to 741741 to reach the Crisis Text Line, and she offers, once, to help you reach someone you trust. If you say no, she respects that and keeps talking with you about whatever you need. Behind this, a safety check runs on each message you send to Lexi. It looks only for signs that someone may be at risk of self-harm, and its only job is to help Lexi respond with care. It is a screen, not a diagnosis, and it has limits: some ways of expressing pain, especially indirect ones, can pass through it. Lexi's supportive behavior does not depend on it, and if our AI provider is ever unreachable, the crisis lines still appear. What this system never does: it never stores a record that a heavy moment happened, never adds anything to your Journey record or anything you share, never scores you, and never tells anyone, not a caseworker, not an organization, not your trusted person, unless you choose to reach out yourself. If you have saved a person you trust in Settings, Lexi can offer to help you contact them, and with your separate permission, offer a first draft of a message. You see every word, you can change all of it, and nothing sends unless you send it. Lexi is not a crisis service, a therapist, or a substitute for one. She is built to make sure that in a hard moment, the right doors are in front of you, including the door to a real person you chose. 5. AI Limitations and Error AI systems, including the systems used in this Platform, can make mistakes. Known limitations include: • AI may generate plausible-sounding but inaccurate information ("hallucinations") • AI may reflect biases present in its training data • AI-generated resource recommendations may include outdated, incorrect, or unavailable programs • AI may misinterpret nuanced or ambiguous inputs • AI-generated plans are generalized and may not account for your full individual circumstances We continuously work to improve the accuracy and reliability of AI-generated content on the Platform. We encourage you to verify all resource information independently and to seek professional advice for important decisions. 6. Automated Decision-Making The Platform uses automated systems to analyze your assessment responses and generate support plans. This constitutes automated decision-making as defined under applicable privacy laws including the CCPA and similar state laws. You have the right to request human review of any AI-generated recommendation that you believe is inaccurate or that has significantly affected you. To request human review, contact shane@aieyu.ai. We do not use automated decision-making to deny you access to services, make eligibility determinations for government benefits, or produce legal or similarly significant effects without the opportunity for human review. 7. AI Provider Disclosure We use OpenAI as our exclusive AI provider for U Navigate by AIEYU. We do not currently use other AI providers such as Anthropic, Google, or Microsoft. If we add additional providers in the future, we will update this disclosure. What we send to OpenAI: • Messages you send to Lexi (the chat feature) • Information you enter on intake assessments and questionnaires • Voice recordings of you speaking to Lexi (transcribed via OpenAI's Whisper model) • Documents you upload, when referenced in a chat conversation • Case narratives, for users with admin access to the case review feature • For wellbeing insights: numeric and category signals derived from your check-ins (average mood score, mood trend, most recent mood label, top adjective and top influence) — these are aggregated values, not your reflection notes themselves • Each message you send to Lexi is also sent to OpenAI's moderation service, which screens it for signs of self-harm risk (see "How Lexi responds when things are heavy" in the AI Disclosure). The result shapes Lexi's reply and is never stored. • If you have saved a trusted contact, their name (never their phone number) is included with requests to OpenAI so Lexi knows who they are • If you have turned on "Let Lexi offer a first draft" and ask for one, your person's name is sent to OpenAI to write that draft. The draft is shown to you and is never stored. • When the Platform suggests a title for a journal entry, the beginning of that entry (up to about 2,000 characters) is sent to generate the title What stays on our servers (NOT sent to OpenAI): • Your wellbeing reflection notes — the words you write in reflections do not leave our servers; OpenAI only receives the derived signals listed above to generate the insight text • Your profile demographics (race, sex, identity fields, contact information) • Your password, payment information, and login credentials Photos are read differently than files. A PDF or Word file already has the words inside it, so we pull them out exactly. A photo has no words in it, only a picture of words, so the words have to be guessed at from the shapes. Most of the time the guess is right. Sometimes it is wrong, and a wrong guess can look just as confident as a right one — a date, a case number, or a name can come out changed without anything looking off. So we keep your photo, always, and treat the text as a helper rather than as the document. Before you rely on something the text says, check it against the picture. If you have the file itself rather than a photo of it, upload that instead. It will be read exactly. What OpenAI does with the data: • OpenAI keeps content sent to its API in abuse-monitoring logs for up to 30 days, then deletes it, per their Standard API terms. It keeps content longer only where the law requires it or where needed to prevent harm. • OpenAI uses these logs only to monitor for abuse of its services. • OpenAI does NOT use this content to train its AI models, per their Standard API terms. These terms apply to our current Standard API account with OpenAI. If our account terms change (for example, to a Zero Data Retention contract or an enterprise agreement), we will update this disclosure. 8. Contact For questions about AI use on the Platform, contact us at: AIEYU Email: shane@aieyu.ai