After backlash, Spotify removes a clause that let Apple use audiobooks from Findaway Voices to develop synthetic voices; Spotify acquired Findaway in June 2022
perhaps in part because they weren't really aware it was happening or what it could lead to. https://twitter.com/... Thanks: @phonesoldier
Context & Ripple Effects
Spotify bought Findaway to become an audiobook producer, distributor, and retailer all at once (the deal reshaped it into an end-to-end audiobook company), with Daniel Ek planning a freemium model while the DOJ reviewed the then-pending acquisition (Ek's audiobook plans and the DOJ review). Findaway Voices is the self-publishing arm of that business, which means its terms govern independent authors' and narrators' catalogs.
The clause now withdrawn sat inside those publisher-facing terms: it granted Apple rights to use Findaway Voices audiobooks to develop synthetic voices — a quiet transfer of training data to the very platform whose App Store rules had already forced Spotify to strip audiobook purchasing from its iOS app (Spotify's iOS compliance move). The backlash was less about one clause than about authors learning their work fed a competitor's voice AI without meaningful awareness or consent.
First-order effects
- Authors and narrators distributing through Findaway Voices no longer grant Apple implicit rights to their recordings for synthetic-voice development; the default terms now exclude that use.
- Spotify removes a reputational liability at the exact moment it is positioning itself as an all-in-one audio company — but it also loses whatever value Apple placed on that catalog access.
Second-order effects
- Apple, cut off from passive access to Findaway Voices material, would need to negotiate explicit licensing deals with publishers or record synthetic-voice training data directly — turning consent into a priced input rather than a contract footnote.
- Other audiobook distributors face pressure to audit their own terms for similar AI-training clauses, since the backlash establishes that authors will treat undisclosed voice-cloning rights as a violation rather than boilerplate.
Third-order effects
- If the pattern holds, back-catalog audio terms become a contested IP frontier: platforms holding large spoken-word libraries either monetize them as licensed AI training corpora under opt-in consent, or defend them against rivals seeking the same — with regulators likely watching how consent is handled given the DOJ's earlier scrutiny of this very acquisition.
- Synthetic-voice economics push toward a two-tier market: catalogs with explicit voice-use rights command premium licensing value, while unencumbered-by-consent catalogs get repriced downward as buyers can no longer assume blanket usage.
The trend: As synthetic voice technology matures, the fine-print terms governing existing audio catalogs are becoming an active licensing battleground where platform defaults get rewritten under public pressure.