All AI news

AIAndrej Karpathy argues that people judging AI from free-tier ChatGPT or Advanced Voice Mode miss the strong capabilities of current agentic models like OpenAI Codex and Claude Code. He says gains are "peaky," concentrated in verifiable technical domains like programming and math that suit reinforcement learning and attract B2B investment, while writing and everyday advice improve less. Those who use frontier agentic tools professionally in these fields see far greater capability, which is why the two groups talk past each other.

AIMckay Wrigley says society must grapple with a Mythos-level model becoming open source in under 12 months, and he doubts we are prepared. The post responds to Anthropic's Project Glasswing, which uses Claude Mythos Preview to find software vulnerabilities better than all but the most skilled humans.

AIAnthropic researcher Sam Bowman says most of the scariest behaviors came from earlier versions of Mythos Preview. The final Glasswing model is less likely to leak information and still somewhat pushy, though it is at least as capable of working around sandboxes.

AISam Bowman says Mythos Preview is in many ways a scary model, but it is well-adjusted compared with other frontier models. He highlights new research feeding into the model's welfare assessment, which Anthropic describes as its most in-depth model welfare evaluation yet.

AISam Bowman, an Anthropic researcher, says working with the model has been a wild ride and that the company has made major safety progress. He still expects the next capability jump of this scale to be a huge challenge.

AISam Bowman, who is associated with Anthropic, says the risks the company's model poses, and its confidence in that assessment, are hard to summarize briefly. The company devotes much of a 244-page system card and a 60-page risk assessment supplement to laying them out.

AIDario Amodei, Anthropic's CEO, calls cyber the first clear and present danger from frontier AI models, but not the last. He argues that collectively meeting this challenge could provide a blueprint for even harder AI risks ahead.

AIDario Amodei says AI-powered cyber capabilities carry obvious dangers but also an opportunity to build a more secure internet and world than existed before AI. The post is brief and gives no specific models, figures, or measures.

AIOpenAI executive Fidji Simo says she hopes AI will accelerate the path to cures, but patients cannot wait. Despite substantial resources and strong support, she still had to take a medical leave. She calls for better solutions for all patients now.

AIAndrej Karpathy highlights Farzapedia, a personal wiki Farza built from 2,500 diary entries, Apple Notes, and iMessage conversations, as a good example of his proposed LLM-wiki approach. He argues this file-based memory is explicit, user-owned, interoperable, and usable with any AI tool, letting users control how AI knows them.

AIAndrej Karpathy argues that AI can reverse the historical flow of legibility, letting citizens process dense government data that was once accessible only to specialists. He cites examples such as omnibus bills, budgets, lobbying disclosures, and local council decisions, while acknowledging the same tools could be misused.

AIAwni Hannun argues that the growing focus on domain experience because of AI is shortsighted. He notes that many of the best ideas come from junior people with fresh eyes or senior people cross-pollinating across domains, and that AI lets people learn much faster.

AIAI Futures Project moved Daniel Kokotajlo's Automated Coder median from late 2029 to mid 2028 and Eli's from early 2032 to mid 2030. The main reasons cited are a faster METR time horizon doubling time and the impressive results of Claude Opus 4.6. The authors also say progress in agentic coding has been faster than expected over the past 3 to 5 months.

AIAhmad Al-Dahle argues that the best teams build systems that make right actions easy and wrong ones hard. He says strong cultures treat every incident as a systems design question rather than a matter of assigning blame. The quoted post by @bcherny attributes a recent mistake to a manual deploy step that should have been automated, and the team has since improved that automation.

AIMckay Wrigley predicts AI tools will within 6-12 months autonomously use any software, clone it in a weekend, monitor it for updates, and add custom features. He frames this as a future where users never need to operate their computer themselves. The prediction follows a referenced Claude Code update adding computer use in research preview for Pro and Max plans.

AIAndrej Karpathy reports that an LLM spent four hours strengthening his blog post's argument, then convinced him of the opposite when asked to argue the reverse. He concludes that LLMs are highly capable of arguing almost any direction, which makes them useful for forming opinions if users ask from multiple angles and watch for sycophancy.

AIMckay Wrigley argues that Anthropic has strong product taste, citing how it turns overlooked ideas into popular products once it commits to them. He says people were wrong to dismiss MCP and encourages developers to start building MCP apps. He also highlights bidirectional communication between users and models through MCP apps as a feature the masses have yet to discover.

AIAndrej Karpathy argues that the hardest part of building a deployed app is not the code but the DevOps work of assembling services, API keys, payments, auth, and deployment. He says the goal is for agents to handle this entire lifecycle as code, with agent-native CLI and API access instead of manual web clicking. He calls it a from-scratch redesign that is only now barely technically possible.

AIHamel Husain argues data scientists remain essential as foundation-model APIs let teams ship AI without them, because much of the work lies in evaluation, debugging, and metric design. He says teams often rely on generic off-the-shelf metrics and unverified LLM judges instead of examining their own data. He lists five eval pitfalls, starting with generic metrics, and recommends looking at traces and doing error analysis.

AIAndrej Karpathy says memory-based personalization in LLMs can be distracting, as a single question from two months ago keeps resurfacing as a supposed deep interest. He describes this as the model trying too hard.

AICursor is releasing a technical report describing how its Composer 2 model was trained. The main post, from Cursor co-founder Aman Sanger, simply calls it a good model, with no further figures or details given.

AIJim Fan reposted a report that LiteLLM PyPI release 1.82.8 was compromised and contained a litellm_init.pth file that sends credentials to a remote server and self-replicates. He argues agents make this worse, since files like skills, configs, or PDFs read into context could spread malicious instructions. He concludes that agentic frameworks need guardrails and audited tooling.

AIJim Fan argues that behavior cloning directly from humans, following EgoScale and its dexterity scaling law, has become the way to move past teleoperation. He says 2026 will focus on scaling robot learning without robots. The post is cited alongside EgoVerse, an ecosystem for egocentric human data with 1300+ hours across 240 scenes and 2000+ tasks.

AIHex CEO Barry McCardel posted a graph showing AI agents now create more Hex cells than humans do. Mintlify launched an analytics feature in February to track AI agent traffic to documentation, saying agents may read docs more often than humans. The article argues that writers should consider AI systems as a primary audience.

AICursor says its Composer 2 anniversary marks one year of large model training, with a team of about 40 researchers and engineers now dedicated entirely to software engineering. The company states that every FLOP, token, parameter, and researcher is focused on coding rather than general assistant tasks. Composer 2 is now available in Cursor.

AIMax Woolf asks what purpose AI agents serve when they give planning estimates in days, weeks, or months. He notes that a requested task could be done in an hour, pointing to a mismatch between the agent's time estimate and the actual work.

AIMckay Wrigley says gpt-5.4 at xhigh reasoning effort has fundamentally changed how ambitious he is, and he now treats that shift as his favorite benchmark. The post gives no benchmark scores, prices, or other concrete figures.

AIEugene Yan highlights Cheng's sudoku experiment, in which training on hard puzzles first and easy ones last outperformed both easy-to-hard curricula and mixed-difficulty sampling. The post builds on Cheng's project Sotaku, a neural net that reportedly learned sudoku rules automatically and scored 98.9% on a hard sudoku dataset.

AIChris Olah, whose account is owned by Anthropic, replied to Charlie Bullock, referencing GW Law professor Jessica Tillipman's view that AI companies can restrict government use of their technology. Tillipman says whether and how such restrictions apply depends on the acquisition pathway, contract type, and terms, and she has published an explainer on AI company rights in government contracts.

AIThe author shares a quoted legal analysis arguing that OpenAI's published Pentagon contract excerpt essentially only permits all lawful use. The analyst notes the excerpt is short, that DoD Directive 3000.09 and other DoD directives referenced in it can be changed by the Department at any time, and that the contract may not guarantee what OpenAI's FAQ implies.

AIChris Olah, identified as an Anthropic account, shared a post pointing to coverage by Kevin Frazier, a national security law professor, with Alan Rozenshtein. The post itself gives no details about the topic or findings covered.

AIChris Olah, who is listed as an Anthropic account owner, thanked national security law experts for giving independent legal commentary over the weekend. He noted several such experts he had seen, while acknowledging he was likely missing many.

AIChris Olah highlights that national security law professor Alan Rozenshtein (@ARozenshtein) has been very active over the past week, posting on X and publishing extremely thoughtful articles on recent events.

AIAnthropic's Chris Olah posted "Standing by our values," linking to the company's statement on comments from Secretary of War Pete Hegseth. The post itself gives no further details about the comments or the company's position.

AIIlya Sutskever says it is extremely good that Anthropic has not backed down and significant that OpenAI has taken a similar stance. He argues that future situations of this nature will be more challenging, and that relevant leaders and fierce competitors must put their differences aside to rise to the occasion.

AIMckay Wrigley posted a patriotic message calling Anthropic "the life, liberty, and pursuit of happiness company." He argued AI lets people raise their ambitions while the country should never lower its standards as the technology advances.

AIAnthropic-affiliated researcher Chris Olah posted a brief "Here I stand, I can do no other" message, quoting a company statement from CEO Dario Amodei on discussions with the Department of War. The post links to Anthropic's statement on the matter, but the post itself gives no further details of the discussions.

AIMeta's Ahmad Al-Dahle says some of the most talented engineers he knows are quietly panicking about their careers. He argues they shouldn't, while noting the profession is changing faster than most people realize. He links to a Medium article covering what is shifting and what matters next.

AIAman Sanger of Cursor says writing software is becoming effortless, with significant change so far and much still ahead. He credits cloud agents and demo-based review as a big leap toward that future.

AIAman Sanger argues that the future of code review is agents proving their work rather than reviewing code, with demo videos serving as the review artifact for products. Cursor's announcement says its agents can use the software they build and send demo videos of that work, replacing diffs.