DaVoice alleges that Perplexity AI stole its wake‑word code and training data, sparking a trade‑secret lawsuit.
DaVoice, a rising voice‑AI startup, has filed a federal lawsuit accusing Perplexity AI of misappropriating its proprietary wake‑word detection code and training data.
Allegations of Trade‑Secret Theft
DaVoice claims that former employees who joined Perplexity AI took confidential source code and large datasets used to train its wake‑word recognition system, then incorporated the material into Perplexity’s own products without permission.
The complaint, filed in the U.S. District Court for the Northern District of California, seeks injunctive relief, damages, and an accounting of any profits derived from the alleged theft.
Background on DaVoice’s Technology
Founded in 2021, DaVoice developed a low‑latency, on‑device wake‑word engine designed to operate without constant cloud connectivity, a feature marketed to privacy‑focused enterprises.
The company’s technology relies on a custom neural network architecture and a curated speech corpus that it says took years to assemble and refine.
Perplexity AI’s Response
Perplexity AI has denied the accusations, stating that its wake‑word capabilities are built from publicly available datasets and its own independent research.
In a brief statement, the company called the lawsuit “unfounded” and indicated it would vigorously defend its position in court.
- Seek a preliminary injunction to halt Perplexity’s use of the disputed code
- Pursue monetary damages for alleged misappropriation
- Request a detailed accounting of any revenue generated from the contested technology
“We will protect our intellectual property and the investments of our investors,” DaVoice’s CEO said in a press release.
The case highlights growing tensions in the AI industry over the ownership of training data and the fine line between open research and proprietary innovation.