Back to News
PolicyAI Understanding briefing

Startup sues Perplexity AI over alleged theft of voice assistant wake‑word technology

A California court filing accuses Perplexity AI of appropriating proprietary wake‑word tech from DaVoice, a small AI startup, raising questions about IP protection in the fast‑growing voice‑assistant market.

4 min readRead the linked source
Source-provided image accompanying Startup sues Perplexity AI over alleged theft of voice assistant wake‑word technology
Source referenceSource recorded
Publisher
news.bloomberglaw.com
Source link
news.bloomberglaw.comhttps://news.bloomberglaw.com/tech-and-telecom-law/startup-claims-perplexity-stole-voice-tech-for-ai-assistants
Source type
Linked source — primary-source status has not been established.
ContextUnderstand this in 60 seconds

Start here

Key terms

Perplexity
A language-model metric measuring how surprised the model is by true next tokens.
Test yourselfAI Ethics Quiz

What happened

A small AI startup, SytteMLLabs (doing business as DaVoice), filed a lawsuit in the U.S. District Court for the Northern District of California alleging that AI, valued at about $20 billion, misappropriated its proprietary wake‑word technology after a promised long‑term partnership. The complaint, heavily redacted, claims Perplexity used the partnership to extract DaVoice’s core “gatekeeper” tech that enables devices to continuously listen for a trigger phrase before activating an AI assistant, then built its own system without compensation. The filing, submitted Thursday, seeks damages and an injunction to stop Perplexity from using the contested technology.

On Thursday, DaVoice submitted a complaint alleging that AI breached a partnership agreement by extracting DaVoice’s proprietary wake‑word detection technology. The lawsuit claims the technology is a "critical gatekeeper" that enables continuous listening for a trigger phrase, a function essential to many AI assistants.

According to the filing, promised a long‑term collaboration but instead used the access to copy and integrate DaVoice’s core algorithms into its own voice‑assistant offering. The complaint seeks monetary damages and an injunction to prevent further use of the disputed technology.

The court document is heavily redacted, limiting public visibility into the technical specifics and the exact terms of the alleged partnership. Bloomberg Law reported the filing but did not provide the full text of the complaint.

Source details: news.bloomberglaw.com ↗

Why it matters

Wake‑word detection is a foundational component of voice‑enabled AI assistants, allowing hands‑free interaction while conserving power and privacy. Control over this technology can give a company a competitive edge in the burgeoning voice‑assistant market, which includes giants like Amazon, Google, and Apple. If the allegations prove true, the case could set a precedent for how larger AI firms acquire and integrate third‑party innovations, potentially tightening due‑diligence and licensing practices. Moreover, the dispute highlights the broader challenge of protecting intellectual property in an industry where rapid integration of niche capabilities is common, and where legal frameworks have struggled to keep pace with AI‑driven product development.

Wake‑word technology underpins the user experience of voice assistants, influencing activation speed, accuracy, and privacy safeguards. Ownership of a robust wake‑word system can differentiate a product in a crowded market.

The case underscores the tension between rapid AI innovation and the protection of smaller firms' intellectual property. A ruling favoring DaVoice could encourage stricter licensing practices and more cautious integration of third‑party tech by large AI platforms.

Legal outcomes may affect investor confidence in AI startups that rely on partnership models for scaling, potentially reshaping funding dynamics in the voice‑AI sector.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

System Requirements:
Best ArchitecturePure RAGRecommended pattern
Hallucination RiskVery LowGrounding efficacy
Update Cost$0 (Vector sync)Ongoing maintenance
Core takeaway: Fine-tuning teaches models how to speak (form, style, syntax); RAG teaches models what to say (verifiable facts). Never use fine-tuning alone for factual memory.
Interactive Concept Check+10 Points
AI Ethics Quiz

Impossibility results in algorithmic fairness (e.g. Kleinberg et al., Chouldechova) show what?

What to watch next

Stakeholders should monitor the court’s rulings on preliminary injunctions, the scope of any damages awarded, and whether opts to settle or contest the claims. The outcome may influence future partnership agreements between AI startups and larger platforms, prompting more explicit licensing terms. Industry observers will also watch for any ripple effects on other voice‑technology providers, as well as potential regulatory scrutiny of IP practices in AI. Finally, the case could spur legislative interest in clarifying IP rights for AI‑related software components.

Whether the court grants a preliminary injunction that forces to halt use of the contested wake‑word system while the case proceeds.

Potential settlement negotiations that could include licensing agreements, royalty payments, or technology transfer arrangements.

Reactions from other AI and voice‑assistant companies, which may revise partnership contracts to include clearer IP safeguards.

Any regulatory commentary or legislative proposals aimed at clarifying IP rights for AI‑related software components.

Related guides & quizzes

AI EthicsAI AgentsAI Models ExplainedTest what you know — try a free AI quizLook up an AI term in our glossaryFollow the AI regulation tracker
Found this useful?