Ad
Skip to content

Maximilian Schreiner

Max is the managing editor of THE DECODER, bringing his background in philosophy to explore questions of consciousness and whether machines truly think or just pretend to.
Read full article about: Tech giants push for priority chip access through new export restrictions

According to the Wall Street Journal, Amazon, Microsoft, and AI startup Anthropic are backing a US law that would further restrict Nvidia's chip exports to China. The proposed Gain AI Act would require semiconductor companies to satisfy US demand first before shipping chips to countries under arms embargos. The law would give tech giants like Amazon and Microsoft priority access to chips.

Nvidia opposes the plan, warning it would create unnecessary market interference. Some government officials question whether the law is even needed, pointing out that the Commerce Department already has the authority to enforce export controls. Meta and Google haven't commented on the proposal. The Gain AI Act could be attached to the defense budget as an amendment.

Comment Source: WSJ
Read full article about: Google releases updated version of Gemini Live

Gemini Live gets its biggest update yet. Google says its voice-based AI is now faster, more expressive, and capable of speaking with different accents. Users can also control the speed of its replies. The new update is also said to improve language learning and pronunciation practice.

The company first teased the update back in August during its Made by Google event.

Read full article about: Google introduces Private AI Compute to protect user data during AI inference

Google is introducing a cloud-based system called Private AI Compute designed to protect user data during AI processing. Jay Yagnik, Google's Vice President of AI Innovation, said the technology runs tasks inside an isolated environment that no one - not even Google - can access.

The system uses Google's own TPUs along with Titanium Intelligence Enclaves for encrypted data handling, building on the company’s existing privacy and security framework.

Early applications appear on Pixel devices, including Magic Cue and the Recorder app, which now supports more languages. The goal is to let Gemini models deliver their full performance without exposing personal data. Google has also published a technical brief outlining the system's architecture and privacy safeguards.

Read full article about: Nvidia wants to turn the data problem in robotics into a compute problem

Nvidia turns to synthetic data to tackle robotics’ biggest challenge: the lack of training data.

"We call this the big data gap in robotics," a Nvidia researcher said at the Physical AI and Robotics Day during GTC Washington. While large language models train on trillions of internet tokens, robot models like Nvidia’s GR00T have access to only a few million hours of teleoperation data, gathered through complex manual effort - and most of it is narrowly task-specific.

Nvidia’s answer is to rethink what it calls the "data pyramid for robotics." At the top sit real-world data - small in quantity and expensive to collect. In the middle lies synthetic data from simulation - theoretically limitless. At the base is unstructured web data. "When synthetic data surpasses the web-scale data, that's when robots can truly learn to become generalized for every task," the team states. With Cosmos and Isaac Sim, Nvidia aims to turn robotics’ data shortage into a compute challenge instead.