All Dispatches

TP-Link Kasa cameras have been found to leak home GPS data via unauthenticated UDP requests, exposing user locations. Firmware updates are available to patch the vulnerability.

A rare glimpse into Amazon's shadow bribery market has been revealed, with sellers being offered access to company employees for a price. The market, which operates on messaging apps like WeChat and Telegram, has been a persistent scourge of Amazon's online store.

Kaiser Permanente nurses claim AI-powered workplace surveillance is compromising patient care and their ability to provide empathetic support. As contract negotiations begin, nurses are pushing back against the use of AI in call centers, citing concerns over automated systems assessing their empathy and tone of voice.

Mozilla's inaugural State of Open Source AI report reveals that open models now power roughly a third of real-world AI usage, but capture only 4% of the revenue. The report highlights the increasing competitiveness of open source AI with proprietary systems, emphasizing its societal and economic benefits.

The Roc team reflects on their Rust-to-Zig rewrite journey, discussing feature parity, hot code loading, and build times. Learn about their experiences and the benefits of using Zig.

OnePlus, a consumer electronics brand, will cease operations in the US and Europe as part of a larger restructuring under its parent company Oppo. Existing devices will continue to receive software updates and security patches, but no new products will be launched in these regions.

Moonshot AI releases Kimi K3, a 2.8-trillion-parameter open model with native multimodality and a Mixture-of-Experts architecture. The model achieves approximately 2.5× better scaling efficiency than its predecessor and is available via Kimi.com, Kimi Work, Kimi Code, and the Kimi API.

The author delves into the complexities of reporting free or used memory in modern operating systems, specifically FreeBSD, and discovers issues with the `btop` tool. A fix is proposed and implemented, providing a more accurate representation of memory usage.

A developer successfully ran Google's Gemma 4 26B model on a 13-year-old Xeon with no GPU, achieving 5 tokens per second decode speed. The journey to get there reveals the state of local LLM inference on non-standard hardware.

A new wave of AI-powered voice fraud is sweeping the nation, leaving law enforcement and financial institutions struggling to keep up. With just three seconds of audio, scammers can create a convincing clone of a person's voice, allowing them to trick victims into handing over sensitive information or money.

Exploring alternatives to run CUDA on non-Nvidia hardware, including HIP, SYCL, and ROCm.

The era of free open-source software is coming to an end. Maintainers are burning out, and companies are exploiting their labor without contributing back. It's time to rethink the economics of open-source software.