Claude Voice Mode Expands to Opus and Sonnet — Anthropic has expanded Claude voice mode to its more capable Opus and Sonnet models, allowing users to move between text, voice, and different models during a conversation. (The Verge)
Amazon Reduces Alexa Costs by Transitioning Models — Amazon is reducing Alexa Plus's reliance on Anthropic models and routing more requests through its own systems to lower the cost of operating the AI-powered assistant. (Business Insider)
Microsoft Launches Enterprise Speech Model to Cut Costs — Microsoft has introduced MAI-Voice-2-Flash, an in-house speech model designed for high-volume voice applications and enterprise contact-center workloads. (Microsoft)
Synthesia Launches Roleplay Sessions — Synthesia has launched interactive Roleplay Sessions in which AI avatars conduct realistic conversations, provide personalized coaching, and assess workplace communication skills. (Synthesia)
Richmond Adopts AI Assistant — The city of Richmond is using an AI assistant to answer and route non-emergency calls using city-approved information, reducing pressure on human dispatchers. (Axios)
Meta Pauses Smart Glasses Cap — Meta has paused its planned subscription test and usage limits for Conversation Focus, an on-device smart-glasses feature that amplifies the voice of the person a user is speaking with. (The Verge)
OpenAI Launches Presence Agent Platform — OpenAI has launched Presence, a managed enterprise platform for deploying governed voice and chat agents across customer-support and internal business workflows. (OpenAI)
OpenAI Brings Desktop Voice Control to ChatGPT — ChatGPT Voice is now available in Work and Codex on desktop, allowing Business users to start tasks, check progress, and coordinate multiple agents through spoken conversations. (OpenAI)
Deepgram Optimizes Speech Recognition for Snapdragon PCs — Deepgram has optimized its Nova-3 speech-to-text model for the Qualcomm Hexagon NPU, enabling private, real-time, on-device speech recognition on Snapdragon-powered PCs. (Deepgram)
Mozilla Introduces Voice-Powered Answers for iOS — Firefox Quick Answers lets iOS users ask questions by voice and receive concise answers with supporting links, while processing raw voice input through on-device speech recognition. (Mozilla)
Papua New Guinea Drafts Laws Against Voice Cloning — Papua New Guinea is preparing legislation to criminalize malicious AI uses including fraudulent voice cloning, deepfakes, digital impersonation, and AI-generated abusive material. (ABC News)
Voice May Define AI's Next Chapter — Industry leaders argue that natural, multilingual voice interaction may become the primary interface for everyday AI use across banking, retail, healthcare, and other services. (The Economic Times)
South African Firms Rush to Deploy Voice Agents — A growing number of South African companies are developing AI voice agents for outbound calls and customer interactions as enterprise adoption accelerates. (News24)
LG Uplus and Ericsson Develop Network-Based Voice AI — LG Uplus and Ericsson have partnered to explore voice AI services embedded directly in telecom networks, reducing dependence on individual apps and devices. (Ericsson)
Google Faces AI Voice Misappropriation Suit — A federal judge ruled that former NPR host David Greene's claims accusing Google of misappropriating his voice for an AI podcasting product can proceed in state court. (Law360)
Apple Fails to Consolidate AI Voice Suits — An Illinois federal court rejected Apple's request to consolidate eight proposed class actions alleging that technology companies used journalists' and voice actors' voices to train AI systems. (Law360)
Apple Wins Appeal in Speech Recognition Patent Dispute — The Federal Circuit vacated and remanded a patent decision after finding errors in the analysis of claims covering a speech-recognition circuit and acoustic-model memory. (IPWatchdog)