AI Policy

Last updated: 11 September 2026. This draft policy is for lawyer review.

Roasty uses AI to help people practice language through conversation, feedback, correction, and text-to-speech responses.

Speech, microphone input, transcripts, lesson context, and generated feedback may be processed by AI and voice providers such as OpenAI and ElevenLabs.

TTS audio files may be generated and stored or cached to provide app features and improve performance.

AI feedback can be wrong, incomplete, or overly confident. Roasty should make uncertainty clear when relevant.

Roasty is for language learning and should not be used for professional advice, emergency decisions, or sensitive personal disclosures.

Roasty's personality is playful and mischievous. Roasts should target the language mistake, not a person's identity or protected traits.

Users should not prompt Roasty to generate harassment, hate, sexual content involving minors, or illegal instructions.

Questions or concerns about AI processing can be sent to privacy@example.com.

We will continue refining AI behavior, provider choices, and data controls as the product approaches launch.