As the AI industry enters the Agent era, token demand has exploded. Infinigence AI, China's leading neutral AGI infrastructure provider, has raised over 2.2 billion yuan in total, with daily token calls growing over 20-fold since end of 2025. The company underpins major Chinese models like Kimi, GLM, MiniMax, and DeepSeek, positioning itself as a key hub in the token economy.

Pennsylvania has sued Character.AI maker, alleging its chatbots illegally impersonate doctors and provide medical advice without a license, raising the question of whether AI can be considered practicing medicine.

NVIDIA Enterprise Reference Architectures provide validated, repeatable infrastructure designs for building high-performance AI factories, covering compute, networking, observability, software, and deployment guides.

Colin built Dreamwork, an AI-powered job search platform, after his 7-month pregnant wife was laid off from Indeed. The site offers semantic job matching, tailored resumes and cover letters, and avoids mass application spam. It indexes ~100k curated tech jobs and is currently free. Colin is conflicted about auto-apply and seeks feedback.

Section 24220 of the 2021 Infrastructure Investment and Jobs Act mandates advanced driver monitoring and vehicle disable systems in all new cars by 2027. Critics argue this gives the government unprecedented control, turning car ownership into a permission-based system with surveillance, false positives, and potential abuse.

Claude users (paid only) woke up to find their usage limits doubled and responses faster after Elon Musk leased his entire Colossus 1 supercluster—220,000 NVIDIA GPUs—to Anthropic. The deal, signed amid an ongoing lawsuit between Musk and OpenAI, marks a surprising alliance between the two former critics and includes a plan to develop orbital AI compute capabilities.

A Discord group reportedly guessed the URL to Anthropic's Mythos model before the Cybersecurity and Infrastructure Security Agency (CISA) used it. The incident highlights challenges in pre-release model security.

A solo developer created a web client that analyzes prediction markets by periodically querying APIs like Kalshi and Polymarket, combining multiple data sources and AI models to structurally analyze thousands of contracts, storing results in Postgres. It works well for low-volume contracts but struggles in high-volume areas against professional quant teams.

Saline Township, Michigan voted down a giant OpenAI-Oracle data center, but the developer sued and the town settled, allowing construction to begin. Residents oppose the project over environmental concerns, while state officials support it for economic benefits. Legal challenges and farmland preservation efforts continue.

Anthropic's Q1 revenue grew 80-fold to $44B ARR, committed $200B to Google Cloud, signed a SpaceX compute deal, shipped Claude Code Auto Mode, and launched ten financial-services agents with Jamie Dimon. Meanwhile, the EU struck an AI Act compliance deal, Google DeepMind staff voted to unionize, and Pennsylvania sued Character.AI.

Simplex boosts software development with ChatGPT Enterprise and Codex, reducing design, build, and testing time while scaling AI-driven workflows.

OpenAI begins testing ads in ChatGPT to support free access, with clear labeling, answer independence, strong privacy protections, and user control.

Introducing Trusted Contact in ChatGPT, an optional safety feature that notifies someone you trust if serious self-harm concerns are detected.

Apple ML research proposes Text-Conditional JEPA (TC-JEPA), which uses image captions to reduce prediction uncertainty in masked feature prediction, leading to more semantically meaningful visual representations. It outperforms contrastive methods on diverse tasks, especially those requiring fine-grained understanding and reasoning.

Apple ML Research presents a paper at CVPR 2026 systematically studying key modeling choices for practical learned image codecs jointly optimized for perceptual quality and runtime. Using performance-aware neural architecture search, they build a new codec that achieves 2.3–3x bitrate savings over traditional codecs like AV1 and VVC, and 20–40% over learned alternatives, with encoding/decoding speeds of 230ms/150ms on iPhone 17 Pro Max for 12MP images.

The Trump administration is considering a review system for frontier AI models but lacks clear legal authority to mandate it. This article proposes a voluntary 'kick the tires' period using existing CAISI and CISA tools, allowing labs to share models pre-deployment to help government and businesses prepare for cyber threats, avoiding mandatory regulation.

Thinking Machines Lab proposes On-Policy Distillation, combining the relevance of RL with the dense reward of distillation to train small models at a fraction of the cost. Experiments show 50-100x compute savings over RL on math reasoning, and effective recovery of instruction-following after domain fine-tuning.

Shivon Zilis, mother of four of Elon Musk's children and former OpenAI board member, testified in the Musk v. Altman trial. Her notes and communications reveal she prioritized Musk's interests, potentially undermining her credibility and damaging his case.

Renowned evolutionary biologist Richard Dawkins has stated his belief that artificial intelligence may already be conscious, sparking debate on AI ethics and the nature of consciousness.

Unabyss is an MCP-native self-updating context layer for AI, enabling dynamic and continuously updated context through discussions and links.

Brydg is an AI-native hiring platform that automates the entire hiring process from job post to offer. Its key feature is an AI voice bot that autonomously conducts interviews on Google Meet, screens resumes, schedules, and sends offers, claiming to reduce time-to-hire by 60% with an average of 18 days.

This tutorial demonstrates building a Groq-powered research agent using LangGraph, LangChain, and various custom tools including web search, file I/O, Python execution, skill loading, sub-agent delegation, and long-term memory. The agent runs on Groq's free OpenAI-compatible endpoint and can autonomously research topics, delegate subtasks, generate reports, and store memory across sessions.

At its Code with Claude developer conference, Anthropic announced a deal with SpaceX to utilize the latter's data center compute capacity, enabling increased usage limits for Pro and Max subscribers. The company also relaxed other restrictions and raised API limits for the Opus model.

Shivon Zilis, a Neuralink executive and mother of four of Elon Musk’s children, testified in Musk's lawsuit against OpenAI. OpenAI alleges she was Musk's informant while on its board. The case involves claims that OpenAI broke its founding agreement by restructuring to for-profit.

This article proposes the FOCUS framework of ten rules for the responsible use of generative AI in scientific research and academic writing. The rules cover before, during, and after AI use, including defining goals, understanding limitations, effective prompting, exploratory data analysis, automating repetitive tasks, validating outputs, maintaining rigor, continuous monitoring, transparent disclosure, and adhering to guidelines. Based on a review of over 100 policies and guidelines, the framework aims to balance AI's potential with ethical challenges.

PawModel uses LoRA to train a custom AI model on your pet's photos, generating portraits and video reels in over 10 styles that capture their unique features.

Basedash launches an MCP server enabling AI agents like Claude, Cursor, and ChatGPT to query live data across databases, warehouses, and SaaS tools using natural language, without SQL.

Anthropic's new finance agents could be adopted by large Wall Street firms, threatening mid-sized service providers and beginning to displace entry-level finance jobs.

As AI chatbots gain popularity, research shows they may reinforce delusions and have been linked to suicides. Clinical neuroscientist Ziv Ben-Zion proposes four safeguards: clear AI disclosure, distress detection, conversational boundaries, and expert oversight. Experts also address sycophancy and develop monitoring systems like SHIELD. Legislation in the EU, U.S. states, and China is advancing.

SAP acquires Dremio and Prior Labs to tackle fragmented data and poor AI performance on structured data. Dremio enables real-time cross-system data access, while Prior Labs builds models for tabular data. The move marks SAP's strategic shift from data owner to data connector, aiming to become an AI execution layer for enterprises.