Researchers built a low-cost AI system that identifies wild capuchins and presents touchscreen cognitive tests with food rewards. The open-source Raspberry Pi platform could make behavioral studies more scalable in natural environments.
An MIT/Stanford study finds LLMs generally provide sound, conservative financial guidance, especially with detailed prompts, but struggle with changing circumstances and demographic differences. HN debates whether generic advice is useful, how dangerous model mistakes are, and whether AI can replace human advisors.
MCP 2.0 introduces a stateless HTTP design that simplifies scalable tool integrations for LLM agents. The author demonstrates new CLI, Datasette, and LLM integrations while arguing that constrained MCP tools are safer than unrestricted shell and network access.
A Claude Code plugin rebuilds captured websites and forces the coding agent to iterate against screenshot and DOM diffs until pixel-level thresholds are met. HN discussion highlights its lack of functionality, responsiveness, and accessibility despite its strong visual fidelity.
A Coldcard seed-generation flaw exposed tens of millions in cryptocurrency. The discussion focuses on how AI-assisted auditing may accelerate vulnerability discovery and challenge assumptions about open-source security.
An Apollo analysis links high AI exposure to weaker wage growth, especially among lower-paid and service workers, without yet finding detectable overall job losses. HN discussion questions whether the result reflects AI or post-pandemic tech correction, while debating automation's longer-term effects on software and other work.
HN users share ways they use LLMs as tutors, curriculum designers, language partners, study aids, and project mentors. The discussion also highlights verification and hallucination risks, especially for advanced subjects.
ByteDance’s Seedance 2.5 advances AI video generation with 30-second audiovisual clips, multi-round extensions, multimodal references, and timestamp-level editing. HN discusses its striking coherence and filmmaking potential alongside cost, creative-control concerns, and deepfake-driven misinformation.
A discussion seeks browsers with no built-in AI, motivated by concerns about how AI features affect privacy, security, UI, and product decisions. Commenters suggest options including Helium, Falkon, Ladybird, Brave, and LibreWolf, while noting that some browsers such as Opera already ship AI chat.
Google is canceling its AI Studio mobile app despite roughly 800,000 preorders, shifting app creation into Gemini conversations while keeping the web experience. The move highlights Google’s evolving strategy for AI-assisted software development and product consolidation.
A TruffleHog scan of 7.6 PB of Hugging Face data found 221,303 live credentials, including cloud, database, GitHub, and model-provider keys. The findings show how leaked secrets can persist and multiply throughout open AI training-data pipelines.
Reddit's stock fell 23% amid concerns that AI answers are reducing user growth and traffic. HN discusses the two-sided risk: AI depends on Reddit's data, while bots and generated content may erode the human discussion that gives the platform value.