Author HubMatthias Bastian

Matthias is the co-founder and publisher of THE DECODER, exploring how AI is fundamentally changing the relationship between humans and computers.

AI and society

Aug 23, 2025Aug 23, 2025

Google downplays AI's environmental impact in new study

AI and society

Aug 17, 2025Aug 17, 2025

Meta's human-like chatbot personas can mislead users and result in real-world harm

AI in practice

Jun 28, 2025Jun 28, 2025

Cloudflare CEO Matthew Prince sees trouble ahead for the open web

AI in practice

Sep 2, 2025Sep 2, 2025

Short

ElevenLabs has updated its AI sound effects model to version 2, adding support for clips up to 30 seconds, seamless looping, and a 48 kHz sampling rate. Sound effects are generated using text prompts, either through the web interface or the API. Looping options are now built into the ElevenLabs Studio audio editor. You can try it out here.

The SFX library has been expanded and reorganized, with improved search, a favorites list, and a remix feature. The SB-1 soundboard now supports the updated model and offers MIDI connectivity.

Elevenlabs

AI in practice

Sep 2, 2025Sep 2, 2025

Short

OpenAI has acquired the analytics startup Statsig and named its founder, Vijaye Raji, as CTO for applications. Raji will report to Fidji Simo and take over technical leadership of ChatGPT and Codex.

Statsig, which focuses on A/B testing and feature management, is expected to speed up OpenAI's development cycles. OpenAI has already been using Statsig internally and now plans to fully integrate the platform. For now, Statsig will remain a separate unit in Seattle, with all employees joining OpenAI. The deal is valued at $1.1 billion, according to Bloomberg.

OpenAI

AI in practice

Sep 2, 2025Sep 2, 2025

Mistral integrates MCP support and memory in Le Chat

AI research

Sep 1, 2025

LLMs struggle with clinical reasoning and are just matching patterns, study finds

AI and society

Aug 31, 2025Aug 31, 2025

Short

WeChat is introducing new rules that require users to label any AI-generated content they share, including videos and public posts. The platform may also add its own visible or invisible labels to content to increase transparency.

When posting on a public WeChat account, users must indicate if any content—whether video, image, or text—was generated by AI and choose the appropriate category, including official/media, news, entertainment, personal opinion/reference only. | Image: Screenshot via WeChat

These changes follow China's government regulation on mandatory labeling of AI-generated content, which takes effect on September 1, 2025. Users who ignore the rules, such as by removing required labels or sharing misleading content, will face penalties, according to WeChat.

via Finance Sina / WeChat Coral Security

AI in practice

Aug 30, 2025Aug 30, 2025

Short

Alibaba has developed a new AI chip, which is currently in testing, designed for a broad range of inference tasks, such as powering the responses from a smartphone voice assistant. The chip is manufactured by a Chinese company and is more versatile than Alibaba's older chips. It is designed for inference, not for training AI models—an area where China's biggest weakness lies compared to the US.

As reported by the Wall Street Journal, Alibaba's new chip is compatible with the Nvidia software platform, meaning engineers can repurpose programs written for Nvidia hardware. The chip helps to fill the void created after Nvidia ran into regulatory barriers restricting sales of its products in China. Alibaba was long one of Nvidia's biggest customers before these restrictions were put in place.

Wall Street Journal