AI Disclosures

SparkyEd uses three AI and AI-adjacent systems to provide its services. Each is described below.

What AI does well in this application

• Asking guiding questions that help students think through problems
• Identifying common misconceptions and addressing them Socratically
• Breaking down multi-step problems into scaffolded instructions
• Providing worked solutions with step-by-step explanations at appropriate moments
• Adapting coaching tone and detail to the student's grade level and subject

Known limitations

• AI can make arithmetic and reasoning errors. Large language models are not calculators. Numeric computations should always be verified independently.
• AI responses may not match your student's specific textbook or curriculum. Method names, notation conventions, and problem-solving approaches vary by curriculum.
• AI cannot see the student's actual handwritten work unless a photo is sent via AirDrop, drag-and-drop, QR-code phone upload, or screen capture.
• AI responses are not reviewed by a human before delivery. Parents should periodically review coaching responses for quality and accuracy.

AI is not a substitute for a teacher

SparkyEd is designed as a supplementary learning tool, not a replacement for qualified human instruction. If your student has persistent difficulty with a concept, please consult their teacher, school tutor, or a professional educational specialist.

The following decisions in the application involve AI:

• All coaching text and Socratic questions are AI-generated (Claude)
• The decision of whether a student's answer is correct is inferred by the AI from the conversation. The application does not have access to an answer key.
• Topic identification (labeling what concept a student is working on) is AI-inferred
• Hint level escalation is triggered by programmatic thresholds (time, wrong attempts), not AI judgment. The content delivered at each level is AI-generated.

The following decisions are not made by AI:

• Hint level thresholds (configured by parent in settings)
• Student profile (name, grade, subjects, entered by parent)
• Session logging preferences (controlled by parent)
• Report generation (triggered by parent)

Several U.S. states have enacted or proposed laws requiring disclosure when consumers interact with AI systems. SparkyEd treats AI disclosure as a mandatory legal requirement.

California

California law requires disclosure when consumers interact with AI in certain contexts. SparkyEd discloses at first launch, in this document, in the Terms of Service, and in the Privacy Policy that all tutoring responses are AI-generated.

Colorado

Colorado's AI Act (SB 24-205) imposes requirements on "high-risk AI systems" used in certain consequential decisions. Educational tutoring AI may be subject to future guidance under this Act. We will monitor developments and update our disclosures as required.

Other states

AI disclosure laws are evolving rapidly at the state level. We are monitoring developments in all jurisdictions and will update this document and our in-product disclosures as new requirements become effective. Parents in any jurisdiction should be aware that AI tutoring responses are machine-generated.

Because tutoring interactions are with an AI system, we recommend parents:

• Review the Parent Dashboard periodically to see which concepts the student is working on and how they are progressing
• Occasionally sit with the student during a session to observe the quality of AI coaching responses
• Encourage the student to flag any response that seems wrong or confusing, and to verify it with their teacher
• Not rely on SparkyEd as the sole source of academic support for difficult concepts

You use Anthropic and ElevenLabs services under your own accounts and subject to their respective terms of service and acceptable use policies. You are responsible for complying with those terms. Key documents to review:

• Anthropic Terms of Service: anthropic.com/legal/aup
• ElevenLabs Terms of Service: elevenlabs.io/terms

We confirm that to our knowledge, educational tutoring software is a permitted use case under Anthropic's and ElevenLabs' current terms. However, you remain responsible for verifying this with your own reading of their current terms, which may change.

If you notice a pattern of incorrect, misleading, or inappropriate AI responses, please contact us at legal@sparkyed.com. Your feedback helps us improve our system prompt design and identify areas where the AI coaching approach needs adjustment.