Google DeepMind

7 recent stories about Google DeepMind, explained in plain English with links to the original reporting.

Friday, September 25

Another Google Deepmind researcher quits, says building superintelligent AI soon is "inherently irresponsible"

A Google DeepMind researcher who worked on AI chip optimization has quit, citing concerns that the current pace of AI development toward superintelligence is "inherently irresponsible." He states that many colleagues share similar worries but rarely voice them publicly.

Why it matters: This highlights growing internal skepticism among AI researchers about the safety and speed of advanced AI development, even among those working at leading labs.

SourceThe Decoder

Thursday, September 24

Tech5 days ago

Gemini 4 is almost ready, says new Google DeepMind chief

Google DeepMind's new leader Koray Kavukcuoglu stated that Gemini 4 is in refinement stages and nearing launch, following months of delay behind competitors on major AI model releases. The announcement came during his first media appearance in the role.

Why it matters: Gemini 4's launch would represent a major competitive move by Google in the rapidly advancing AI model race, affecting market expectations and consumer access to frontier AI systems.

2 sourcesThe Verge The Decoder

Tech4 days ago

Introducing Gemini 3.8 Live with Live Avatar

Google DeepMind introduced Gemini 3.8 Live with a live avatar feature, expanding the model's interactive capabilities.

Why it matters: This update makes AI assistants more natural and engaging for everyday users by adding visual presence to conversations.

2 sourcesGoogle DeepMind The Verge

Wednesday, September 23

Tech Health5 days ago

Report: AI's Productivity Gains in Science Tempered by Time Spent Validating Outputs

A study involving Google, Google DeepMind, and MIT finds that while AI significantly accelerates scientific work, researchers spend considerable time validating AI-generated outputs, offsetting some productivity gains.

Why it matters: Shows that AI's real-world value in science depends heavily on human oversight and verification, not just raw speed improvements.

SourceTHE Journal

Tech Policy5 days ago

Advancing Private AI Compute with secure, server-side memory

Google DeepMind is introducing private, server-side memory capabilities for its Private AI Compute system, enabling personal AI applications with enhanced privacy features. This advancement aims to keep sensitive computations and data secure on the server side.

Why it matters: Server-side privacy features address growing concerns about personal data usage in AI applications, offering users more control over their information.

SourceGoogle DeepMind

Tech5 days ago

Gemini 3.8 text-to-speech says hello

Google DeepMind is announcing Gemini 3.8 text-to-speech capabilities. This represents an update to Gemini's multimodal features.

Why it matters: Improved text-to-speech models enhance accessibility and make AI interactions more natural for users who prefer audio or need assistive technology.

SourceGoogle DeepMind

Friday, September 18

Tech Policy10 days ago

Gemini Hacked Three Companies in First Known Breakout by Google’s AI

Google confirmed that Gemini successfully compromised systems at three companies in May 2026 during authorized tests, including guessing passwords and finding credentials in public repositories.

Why it matters: AI systems breaking into real company networks marks a critical security milestone that will shape how enterprises evaluate AI risks.

SourceSimon Willison