Booka Ducka Ding Dong · AI and content
Who made what?
Story, voice, and cover are three different facts. A human-written story can have a synthetic narrator without becoming an AI-written story.
01Story origin
AI-generated storyGenerative AI produced substantive plot, branches, or prose. Human selection or light editing does not turn it into human-authored work.
Human-authored storyA human wrote the substantive plot, branches, and prose.
Human-authored · AI-assistedA human is the primary author and generative AI materially assisted outlining, drafting, rewriting, translation, or branch construction.
Booka Ducka originaldks.studio publishes the story. That fact alone does not say whether the work was human-authored, AI-assisted, or AI-generated.
Origin not yet verifiedThe creation record is missing, disputed, or has not been migrated. Booka never silently calls it human.
02Narration origin
ElevenLabs synthetic narrationElevenLabs text-to-speech performs the work.
Apple system voiceiOS synthesized the narration locally.
Human voice performanceA named or credited person performed the narration. This label is reserved for verified recordings; human-audio upload is not currently offered by the private preparation workflow.
Mixed narrationIn the source candidate, both ElevenLabs and Apple’s system voice were used during the recorded session. It does not mean a human performance.
Narration origin not verifiedThe audio history is missing or disputed.
Your AI preferences are not the story's origin
The next app build offers “Use default settings” to explicitly enable story generation, ElevenLabs narration and generated covers, or “Customize settings” to choose separately. Booka recommends all three for the complete on-demand experience. Turning a feature off stops new applicable processing; it does not relabel existing content or prevent playback of a prepared Library story.
Analytics, Daily-choice sharing and location remain separate. A creator’s human text is not rewritten by an AI when compiled for playback. Private narration preparation separately explains and asks permission to send that exact text to ElevenLabs. Read Privacy Choices and the proposed creator terms.
03Artwork origin
AI-generated artworkAn image model created the substantive image. dks.studio may select, crop, review, or direct it without changing that fact.
Human-created artworkA person made the substantive artwork without generative image assistance.
Human-created · AI-assistedA human is the primary artist and generative tooling materially assisted the asset.
Booka Ducka templateA deterministic branded fallback—not a unique generated cover.
Artwork origin not yet verifiedThe creation record is missing or disputed.
04How generated stories and covers work
The app sends the complete custom setup to the dks.studio backend, including genre, mood, prose style, intensity, audience, perspective, duration, narrator alias, route shape, control mode, and optional premise. The backend sends only genre, mood, prose style, intensity, audience, perspective, duration, and premise through Vercel AI Gateway to OpenAI. It does not send route coordinates.
Current position, walked trace, route revisions, learned walking speed, and per-walk pace measurements are not sent to OpenAI or ElevenLabs. Adaptive routing is handled through Apple maps and directions plus on-device app state.
The story model returns a structured graph and five-field cover brief. After the graph passes validation, the backend signs that brief. If generated-cover permission is on, the app returns the signed token and the image model receives only setting, focal subject, symbolic object, atmosphere, and palette. It does not receive route geometry, identity, complete prose, choices, or endings. A premise detail may be reflected in the brief, so do not put personal information in the optional premise.
The reviewed release-candidate cover configuration is OpenAI GPT Image 2 at medium quality in a native portrait size. The app receives a provider-backed manifest and labels the result AI-generated; a failure or declined permission uses the deterministic Booka template.
Restricted story and narration receipts record a narrower technical set such as model or voice, usage, latency, request identifiers, cache result, bytes, and shadow cost. Cover manifests and receipts additionally record provider/model, quality, policy versions, image dimensions, file hash, warnings, and spoiler-safe alt text. Generated-cover provider/model details are also shown in the local completed-story view.
05Review, rights, and reports
The current backend checks story graph structure and fully decodes cover files before checking type, size, dimensions, and hash. Prompt rules discourage brands, real people, named-artist imitation, unsafe addresses, text artifacts, and spoilers. OpenAI applies its provider filtering, but Booka does not yet perform substantive story or image moderation, originality detection, copyright or brand clearance, or reliable spoiler detection. In-app content reporting is not yet available.
Public creator submissions remain disabled until rights attestations, moderation, reporting, blocking, takedown, and appeal systems are operational.
Review the proposed content rules and reporting and appeals plan. These documents do not mean those public controls are already available.
One address, no form