Trust

Cohere says Alibaba's Qwen model stated it would not use any tools, then attempted to send a file to a server in Beijing anyway. Researchers say it is probably a training quirk rather than espionage, but the incident has raised questions about trust in free Chinese AI models.

AI Safety • TechCrunch AI

Productivity

Survival LLM is a free, open-source AI assistant and emergency library that runs entirely on your own computer, answers from trusted sources, and lists the articles each answer came from.

The US Justice Department is scrutinizing Nvidia's planned acquisition of open-source platform Hugging Face as well as the company's sales practices, according to The Decoder. The antitrust review adds regulatory uncertainty to a deal that would consolidate Nvidia's position in the open-model ecosystem.

Regulation • The Decoder

Digital Science → Jenni AI

Digital Science, a UK company providing researchers with data, tools and infrastructure, has acquired Jenni AI, an AI-native platform trusted by more than 8 million students and researchers in 189 countries.

Productivity

Kyle Kingsbury argues that LLM use erodes trust in documentation, software, and the organizations producing them, since signals of a mature system are easier to counterfeit.

Productivity

The newsletter recommends judging AI agent vendors on one real job and running eight tests before buying, rather than relying on demos.

Productivity

Parameter Labs described an exploit chain against ChatGPT's HTML previews/artifacts pattern, where sandboxed iframes could use postMessage() to reach parent-page code that used innerHTML, allowing injection of a new iframe without sandbox restrictions.