# WiseBoard AI Privacy, Safety, and Instructional Evidence Pack **Audience:** District technology, privacy, curriculum, and procurement teams **Last reviewed:** August 29, 2026 **Purpose:** A concise starting point for evaluating WiseBoard’s AI-enabled classroom features and the evidence attached to instructional recommendations. ## How to read this packet This packet uses three labels: - **Implemented product behavior** — a behavior available in the current WiseBoard product. - **Published documentation** — a linked policy, technical overview, or legal artifact that should be reviewed alongside this packet. - **District configuration or review** — a control that depends on local district choices, policy, or a formal contracting process. This packet is not a certification, a substitute for a district security review, or a guarantee that a recommendation caused a student outcome. ## District procurement quick read Use the status labels below when copying an answer into a procurement response: - **Share now** — supported by a current public policy or implemented product behavior. - **Send supporting artifact** — share the linked document so the district can review the underlying evidence. - **Confirm with owner** — do not turn the item into a contractual promise until the named owner confirms the current wording. | District question | Safe current answer | Status | |---|---|---| | How is data used? | WiseBoard uses district-provided information to provide the requested educational service. Student data is not sold, used for advertising, or used to build commercial profiles. | **Share now** | | Is customer data used to train models? | The current AI Usage Disclosure says WiseBoard uses OpenAI’s GPT-4o family through API endpoints, applies server-side minimization, and does not use customer prompts or completions to train external models. | **Share now** | | How long is data kept? | Retention depends on the record type. The Privacy Policy describes account/service data as active-account data plus 30 days after deletion, with encrypted backups retained for 30 days. Pip transcripts and safety events have separate windows in the Student AI Safety page. | **Share now** | | Is a person accountable for AI output? | Yes. AI output is a draft for educator review. The instructional action loop does not autonomously grade, change records, assign interventions, or message families. | **Share now** | | How are students protected? | Pip’s Study Hall is opt-in per class and off by default, with age-tier controls, teacher and admin controls, distress detection, trusted-adult redirects, transcript visibility, and parent opt-out. Pip is not a counselor or substitute for an adult. | **Share now** | | Is WiseBoard accessible? | WiseBoard describes itself as partially conformant with WCAG 2.2 Level AA and publishes VPATs, known limitations, and alternative paths. Share those artifacts rather than promising full conformance for every surface. | **Send supporting artifact** | | What support is included? | Agree on feature enablement, roles, roster/identity setup, local retention policy, family-notice steps, and support expectations in the district proposal, pilot agreement, or DPA. | **Send supporting artifact** | | What is the incident-notification SLA? | The Trust Center and sales security summary currently describe different timings. The trust owner must reconcile the wording before sales quotes a number. | **Confirm with owner** | ### Current gaps and owners - **Trust / legal:** reconcile incident-notification timing across the Trust Center and sales security summary. - **Sales operations:** attach state addenda, the executed DPA language, and requested security questionnaires to the deal record. - **Finance / operations:** provide the current W-9, ACH instructions, and certificate of insurance when requested. - **Product / AI operations:** reconfirm the active model provider and retention terms whenever a provider, endpoint, or AI feature changes. - **Implementation owner:** document enabled features, staff roles, family-notice steps, local retention policy, and support expectations for each rollout. ## AI use and teacher control **Implemented product behavior** - WiseBoard’s instructional action loop presents classroom signals as recommendations for a teacher to review. - Teachers can approve, dismiss, or correct recommendations. Recommendation actions do not autonomously grade students, alter student records, deliver interventions, or message families. - Each decision stores a snapshot of the recommendation’s visible evidence, source signals, scope, freshness, confidence, limitation statement, and teacher note where supplied. - Recommendations expire and must be refreshed before a teacher can act on stale evidence. - Teachers can review their prior approvals and dismissals, print a coaching summary, and export the authorized history currently in view. **Published documentation** - [AI Usage Disclosure](/legal/ai-usage) describes WiseBoard AI features, data handling, and organization-level controls. - [Student AI Safety](/legal/student-ai-safety) describes the safeguards for Pip’s Study Hall. **District configuration or review** - Districts should review which AI features are enabled for their organization and confirm that local instructional, retention, and family-communication policies are reflected in their rollout plan. ## Recommendation evidence and outcomes **Implemented product behavior** - Recommendation evidence identifies the source signals used to create the prompt and the time/freshness context shown to the teacher. - A recommendation includes a limitation statement: it is a descriptive signal from recent WiseBoard activity, not a causal conclusion. - Outcome checks compare available classroom-level measures before and after an approved action only when sufficient follow-up data exists. - The product does not state that an action caused a measured change. Missing or insufficient data is shown as unavailable rather than replaced with a score. **District configuration or review** - Use exported decision history as a discussion artifact in coaching or PLC conversations. It should be interpreted alongside teacher context, not used as a high-stakes evaluation record. ## Privacy, security, and access **Published documentation** - [Privacy Policy](/privacy) - [Security Overview](/trust/security-overview.md) - [Subprocessor List](/trust/subprocessors.md) - [Subprocessor Change Notification Policy](/trust/subprocessor-change-policy.md) - [SDPC National DPA template](/trust/dpa-template.pdf) - [Incident Response Runbook](/trust/incident-response.md) **Implemented product behavior** - Instructional decision history is scoped to the authenticated teacher’s authorized data. - The admin governance view is protected by WiseBoard’s existing administrator check. - CSV export is generated from the teacher-authorized decision history currently displayed; it does not add cross-classroom or organization-wide access. ## Safety boundaries WiseBoard does not use the instructional action loop to: - make autonomous grading, placement, discipline, or intervention decisions; - change a student record without a teacher-initiated workflow; - send a family message; - treat a descriptive signal as a diagnosis or causal finding. Teachers and districts remain responsible for reviewing recommendations, applying local policy, and deciding whether an instructional action is appropriate. ## Suggested district review questions 1. Which WiseBoard AI features will be enabled for this organization? 2. Which staff roles should access the decision-history export and governance view? 3. How will instructional teams distinguish descriptive evidence from causal claims? 4. Which local policies govern AI-assisted instruction, student support, retention, and family communication? 5. Which linked legal, security, accessibility, and data-processing documents require a formal review? ## Contact For DPA execution, vendor-security questions, subprocessor notices, or other trust questions, contact [info@mywiseboardclassroom.com](mailto:info@mywiseboardclassroom.com).