Google has introduced new Gemini Live models and Gemini 3.5 Transcribe for developers, enhancing the creation of real-time voice applications. These models are accessible via the Gemini API and Google AI Studio, enabling more intelligent conversational experiences.

Specifically, Gemini 3.8 Live and 3.8 Live Extended Thinking are native speech-to-speech models capable of performing tasks while maintaining dialogue, with the latter offering deeper reasoning for complex requests. Gemini 3.5 Transcribe provides highly precise speech-to-text transcription across 85+ languages, achieving an average Word Error Rate (WER) of 4.0% for streaming and 2.6% for non-streaming.

Key capabilities include asynchronous function calling, grounding dialogue in live visual inputs, accurate alphanumeric parsing, multilingual support for over 97 languages, and seamless integration of real-time audio with structured data. All AI-generated audio is also watermarked with SynthID to help prevent misinformation.

👉 Read the full article on Google Blog

Google has introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking to make voice interactions more natural, fluid, and intelligent. These models are capable of handling complex reasoning, real-time visual context, and background task execution without interrupting the conversation.

Gemini 3.8 Live Extended Thinking provides enterprise-grade task completion and intelligence, securing the #1 overall spot on Artificial Analysis' Speech to Speech Quality Index with a score of 82.6. It also leads in agentic task completion with 68.6% on τ-Voice and 35.1% on Sierra's τ-Voice-banking benchmark.

These models process visual inputs in near real-time and automatically detect and transition between 97 supported languages mid-conversation. They execute tools and API calls in the background while continuing the conversation, helping users tackle complex tasks using just their voice.

👉 Read the full article on Google Blog

Major AI companies, including OpenAI, Anthropic, and Google DeepMind, have been collaborating on AI safety issues due to growing concerns about the trajectory and pace of AI development. OpenAI's global head of policy, Chris Lehane, stated that the top AI labs are working together to "prioritize safety."

This collaboration follows an essay by Anthropic CEO Dario Amodei, supported by OpenAI CEO Sam Altman, warning of potentially catastrophic outcomes from unchecked AI development and calling for a slowdown. However, concerns have arisen that an industry-wide coordinated pause could amount to "output restriction," violating antitrust law, specifically the Sherman Act.

Some AI startup executives, such as Cohere CEO Aidan Gomez, portray these calls for safety as a covert anti-competitive scheme where dominant AI labs define the rules and safety standards. Early remarks from the Trump administration also suggest a trickier path forward for any waiver-assisted collaboration.

👉 Read the full article on Gizmodo