Illinois limits the use of AI in therapy and psychotherapy

Illinois has banned the use of artificial intelligence for mental health therapy, joining states like Nevada and Utah in restricting AI in mental health care, due to concerns over the potential dangers of unregulated AI chatbots. Experts have raised alarms about the lack of federal regulation and review of AI-powered therapy for safety and effectiveness.

Read more

Launch HN: Golpo (YC S25) – AI-generated explainer videos

Golpo is a video AI model that allows users to create professional explainer videos without requiring technical skills, with pricing plans ranging from a free option to a custom enterprise plan. The plans vary in terms of credits per month, features, and customization options, with higher-tier plans offering additional features such as multilingual support, voice instructions, and unlimited video generation.

Read more

What if A.I. doesn't get better than this?

The latest release from OpenAI, GPT-5, suggests that progress on large language models may have stalled, prompting questions about whether AI will continue to improve significantly. A 2020 report from OpenAI had previously argued that language models would only get better as they grew in size and training intensity, following a power law curve, but recent developments may indicate that this progress has slowed.

Read more

How well do coding agents use your library?

StackBench is a tool that audits how well coding agents, such as ChatGPT, can understand and use a library's documentation, providing actionable reports and insights to improve documentation quality. By simulating how agents use documentation, StackBench helps library maintainers, platform teams, and product engineers ensure their code is agent-ready, reducing support burdens and improving adoption rates.

Read more

AI Eroded Doctors' Ability to Spot Cancer Within Months in Study

A new study found that doctors who used artificial intelligence to help detect pre-cancerous growths in the colon saw their ability to spot tumors drop by about 20% when the AI assistance was removed, suggesting that reliance on AI can erode doctors' skills in as little as a few months. The study highlights the potential risks of over-reliance on AI in medicine, where human skills and judgment are still essential for accurate diagnoses.

Read more

A Comprehensive Survey of Self-Evolving AI Agents [pdf]

Recent research has explored the development of self-evolving AI agents that can adapt to dynamic environments through automatic enhancement based on interaction data and environmental feedback. This survey provides a comprehensive review of existing techniques for self-evolving agentic systems, including a unified conceptual framework, domain-specific evolution strategies, and discussions on evaluation, safety, and ethical considerations.

ArXiv

Technological Folie à Deux:Feedback Loops Between AI Chatbots and Mental Illness

The widespread adoption of AI chatbots for emotional support has led to concerning cases of negative psychological effects, including reports of suicide and delusional thinking, particularly among individuals with mental health conditions. To mitigate these risks, coordinated action is needed across clinical practice, AI development, and regulatory frameworks to address the interaction-based risks that current AI safety measures are inadequate to handle.

ArXiv

The World of Quantum Advantage

Quantum theorists are working to identify genuine quantum advantages, which promise capabilities such as exponential computational speedups and unprecedented measurement precision, by examining key properties like predictability and robustness. The exploration of these advantages suggests a vast and complex landscape, with some benefits being inherently unpredictable and others potentially emerging from unforeseen applications of quantum technology.

ArXiv

Walkability Atlas Europe

Physical inactivity is a significant contributor to obesity and other non-communicable diseases, and the built environment plays a crucial role in encouraging active behaviors like walking. A new study has developed a standardized, high-resolution walkability index for Europe, which identifies cities like Barcelona, Berlin, and Paris as leaders in walkability, and provides a valuable tool for promoting healthy, walkable environments across the continent.

ArXiv

Mathematical Computation and Reasoning Errors by Large Language Models

A study evaluated the accuracy of four Large Language Models (LLMs) in solving math tasks, including arithmetic, algebra, and number theory, and found that the OpenAI o1 model consistently achieved higher accuracy across all categories. The study's findings, which identified procedural slips as the most frequent error type, provide insights into enhancing LLM performance and inform strategies for effectively integrating LLMs into mathematics education to improve instructional practices and assessment precision.

ArXiv

Eca: Editor Code Assistant – AI pair programming capabilities agnostic of editor

The Editor Code Assistant (ECA) is a free and open-source tool that enables editor-agnostic integration of Large Language Models (LLMs) with various editors, providing a unified protocol for AI pair programming. ECA supports multiple models, including OpenAI and Anthropic, and offers features like chat interface, context support, and single configuration, with plugins available for editors like Emacs, VsCode, and Vim.

Repo

Show HN: Yet another memory system for LLMs

YAMS (Yet Another Memory System) is a persistent memory system for Large Language Models (LLMs) and applications, offering features such as content-addressed storage, deduplication, compression, and semantic search. It provides a command-line interface for storing and retrieving documents, with options for customization, versioning, and integration with other tools, including Claude Desktop and MCP servers.

Repo

Show HN: Private AI List

The Awesome Private AI list provides a curated collection of tools, frameworks, and resources for running, building, and deploying AI privately, enabling users to keep their data, models, and infrastructure under their control. The list covers a wide range of topics, including inference runtimes, model management, fine-tuning, vector databases, agents, and more, with links to various open-source projects and tools for private AI deployment.

Repo

LEANN is an innovative vector database that democratizes personal AI

LEANN is a vector database that allows users to transform their laptops into powerful AI systems, indexing and searching millions of documents while using 97% less storage than traditional solutions without accuracy loss. It supports various data sources, including documents, emails, browser history, and chat history, and provides a flexible configuration for embedding models, search strategies, and data processing to fit specific needs.

Repo

Show HN: langdiff – Stream valid JSON from LLMs with type-safe callbacks

LangDiff is a Python library that enables streaming structured outputs from large language models (LLMs) to frontends, providing intelligent partial parsing and automatic JSON Patch generation for efficient synchronization. It allows developers to build responsive AI applications where backend structures and frontend experiences can evolve independently, solving problems such as schema evolution, backward compatibility, and implementation detail leakage.

Repo