AI Team

A new tool called Pi pod lets developers run the Pi coding agent inside sandboxed environments on their own servers, giving teams control over where AI coding assistants execute code. The project was shared on Hacker News and drew around 120 upvotes, placing it near the top of the site, with developers discussing self-hosting options and sandbox isolation for AI agents.

The MLC team has introduced TIRx Harness, an open compiler harness aimed at agentic GPU programming, announced in a blog post by contributor Jinhong Yi. The tool is designed to let AI agents generate and compile GPU kernels through an open compiler stack. Early reaction online is small but positive, with interest centred on how compiler infrastructure can support agent-driven code generation for GPU workloads.

A newly shared discussion explores how creative and technical teams can use AI to scale their intent, maintain quality standards, and preserve artistic vision as work expands. The conversation reflects a wider debate in creative and engineering circles about whether AI tools can amplify human judgment rather than replace it, and how to keep craft central to scaled production.

Anthropic says a widely used open-weight Chinese AI model demonstrates what it calls Mythos-class hacking capabilities in a new frontier red-teaming report. The company's safety testing also found weak safeguards on openly released AI models, raising concerns that such systems could be repurposed for offensive cyber operations. The findings have intensified debate over the risks of releasing powerful AI weights openly versus the security practices of frontier labs.

A US start-up backed by Nvidia has unveiled a new open-weight AI model, positioning itself as an American answer to Chinese companies that have taken the lead in releasing openly available large language models. Chinese firms such as DeepSeek and Alibaba's Qwen team have gained global attention by publishing powerful models freely, prompting US developers and investors to push for competitive open-source alternatives.

A new survey by financial planning platform Datarails finds that finance teams now spend about a quarter of their working week fact-checking output generated by artificial intelligence tools. The finding highlights a growing hidden cost of AI adoption in corporate finance, where efficiency gains from automation are partly offset by the time staff must devote to verifying accuracy before numbers are used in reports and decisions.

Researchers at ThreatDown have analyzed the new Carbonato botnet, which compromises poorly secured Docker hosts. The malware targets systems where the Docker API is exposed without authentication on port 2375, taking over containers and using AI-generated code in its operations. Security teams are being urged to lock down exposed Docker APIs and review firewall rules to prevent infections.

A fashion house celebrating its 110th anniversary is working to build one of the most AI-fluent teams in the industry, according to a report by Mexico Business News. The story highlights how a heritage brand is integrating artificial intelligence skills into its workforce, a topic drawing attention as legacy fashion companies race to modernize operations and keep pace with technology-driven change across the sector.

Google is restricting its bug bounty program covering open source projects after a surge of low-quality, automated vulnerability reports flooded its review process. The company says AI-generated submissions, many invalid or low-value, have overwhelmed security teams, forcing it to narrow eligibility and tighten criteria for rewards. Security researchers are debating the move, with some warning that legitimate findings may now be overlooked as programs tighten rules across the industry.