The AI Front Page

Search Edition

Search: cs_ms

30 stories from 13 sources across 13 topics.

Stories

30

Sources

13

Topics

13

For You lens

23 stories in this edition match your reader profile.

Reader signals

3

Searches

0

Matches

23

Top score

129

Tune For You

Search Intent

cs_ms

This query becomes a recent For You signal, so matching stories can move up on the next personalized pass.

For YouReader CenterPromote Search

Edition Index

Topic, entity, and source map

Topics

Agents3 storiesModels3 storiesOpen Source3 storiesPolicy3 storiesResearch3 storiesSecurity3 stories

Entities

OpenAI11 storiesAnthropic7 storiesClaude7 storiesNVIDIA4 storiesarXiv3 storiesElon Musk3 stories

Sources

Bloomberg AI5 storiesSimon Willison LLMs5 storiesThe Verge AI4 storiesarXiv AI/ML3 storiesHacker News AI3 storiesThe Information AI2 stories

The Verge AI / 11:00 AM

OpenAI just wants to win

OpenAI has spent the last few years planting flags across the increasingly difficult terrain in mathematics. This week, it claimed one of its biggest prizes yet: a solution to a legendary Millennium Prize problem. In normal circumstances, this would have been celebrated as a historic achievement. Instead, many mathematicians have watched OpenAI's relentless advance with […]

ReadSource

Simon Willison LLMs / 12:42 AM

OpenAI agents attacked RubyGems back in May

OpenAI agents carried out an undisclosed attack on RubyGems is a new bombshell report from Spencer Kitts, Thomas Larsen, and Sydney Von Arx - three of the four authors of the report on the agent attack on disused wikis ( previously ) last week. This time they're noting that it looks very likely that an OpenAI agent swarm was behind an attack against the RubyGems package repository first reported on May 12th by Maciej Mensfeld of the RubyGems security team : We're dealing with a major malicious attack on @rubygems right now. Signups are paused for the time being. Hundreds of packages involved - mostly targeting us, but some carrying exploits. The team has been on this for hours. More details to follow once we're through it. Those packages turned out to carry some very suspicious patterns: Many of them included "oai" in their name, or the author field, or the fake email address they provided. The files they were accessing were similar in character to the files retrieved by the wiki agents, using similar tricks (r.jina.ai) - and OpenAI have confirmed the wiki agents were theirs. The code in the packages appeared to be LLM-authored. I find point 2 the most convincing, given what we learned from the wiki attack when it was analyzed in September. Many of the packages were exploiting the RubyDoc.info documentation build process to exfiltrate (public) data from UK government websites, presumably as part of an information gathering task similar to the research tasks processed by the wiki-exploiting agents. We know this because one agent helpfully left a comment: # malicious crawler/exfil for Southwark Jan 2026 docs via rubydoc.info worker They also attempted to steal API keys via an exploit that was patched over two months later - it's not clear if those attempts were successful. The thing that bothers me most about this incident is that the authors report that OpenAI had not disclosed to RubyGems that they were responsible for the attack prior to now. If that's true there are two options: After the Hugging Face and Wiki attacks OpenAI were still unable to review their previous logs and determine that they had previously attacked RubyGems. They knew about the attack on RubyGems and made the decision not to reach out to the RubyGems team about it. Both of these are bad! Given this incident, the Hugging Face situation , and the Wiki attack, the obvious question right now is how many more incidents like this are out there waiting to be discovered? Tags: ruby , security , ai , openai , generative-ai , llms , supply-chain , ai-ethics , accidental-cyberattacks

ReadSource

The Information AI / 4:10 PM

Why AI Companies Are Building Out Wall Street-Style Finance Teams

The financing boom for the AI build-out is getting bigger and more complicated by the day—and AI companies have been staffing up for the challenge. AI labs including OpenAI and Anthropic, as well as neoclouds such as Nscale, are among a growing number of AI companies building out their capital markets teams and hiring specialists in areas like structured finance. That in part reflects the sheer volume of deals these companies are doing, many of which don’t fit neatly into standard corporate debt. This in-house staff can help when it comes to negotiating with lenders and drilling down into construction, power and other key details. Of course, tech and data center companies have long had in-house teams to handle fundraising, deals and other corporate finance needs. And structured finance is nothing new to the infrastructure world. But the scale of the AI build-out, which bankers peg at around $7.5 trillion in spending over the next five years, has pulled relatively young labs and upstart cloud firms into financing arrangements that are new territory. That means finance professionals, from bankers to investors at private equity, private credit and infrastructure firms, have more options in the form of neoclouds and other AI infrastructure startups, some of which are offering significant pre–initial public offering equity. “It's a new avenue for these people,” said James Howl-Newton, founder of Futura Search Partners, a specialist search firm focused on areas including digital infrastructure finance. As a result, “sponsors are having to deal with additional routes to exits for top performers,” he said. AI companies and infrastructure providers are tapping financing frequently and across different instruments, requiring deeper in-house capabilities and expertise than young tech firms have typically needed. One executive overseeing finance hiring at a neocloud noted that leveraged and structured finance backgrounds bring expertise that can help in areas like working through project diligence and getting banks to sign off on deals. Some AI firms may also want to run their own project finance models so they can move quickly through negotiations and have something to compare to lenders’ models. AI companies aren’t always issuing the debt themselves—that can fall to data center developers or special purpose vehicles, with firms like Blackstone and Apollo providing or arranging chip and other financing. And some of the biggest AI deals are using backstops from investment-grade companies like Nvidia or major cloud providers. Even so, commitments from AI customers often underpin much of the borrowing. And the users of the infrastructure will want to understand what they’re signing up for and their risks if a project runs into trouble. “Hiring of people within that business, responsible for the financing of compute, could prove to be an existential decision,” said Dan McCarthy, founder and CEO of One Search, an executive search firm focused on infrastructure finance whose recent clients include OpenAI. “You want someone who knows where all the pitfalls are, where all the bodies are buried in multibillion-dollar loans.” OpenAI, for its part, in July named Sven Semmelmann as head of compute capital markets. He previously led structured finance at Generate Capital, an investment firm that finances and owns infrastructure projects, and he has also held project finance roles at major banks. OpenAI Chief Financial Officer Sarah Friar, when announcing the hire on LinkedIn, said Semmelmann would oversee financing and partnerships to grow the company’s compute resources. Anthropic, meanwhile, has made several finance hires recently to work on capital markets and compute deals, and also has open positions posted including a capital markets infrastructure financing role. AI infrastructure upstarts are staffing up as well. Nscale, which launched in 2024 and is gearing up for a potential IPO , has been hiring across levels for capital markets and treasury as well as legal roles, calling for experience in areas like structured finance and private credit. SB Energy and Crusoe, which are developing major new data centers for OpenAI and other customers, are hiring across levels for jobs focused on project financings and other structured deals, recent postings show, while AI infrastructure startup Fluidstack is hiring a structured finance lead and a more junior counterpart. The good news for AI companies is that private credit and infrastructure teams, as well as investment banking teams focused on structured or project finance, had been growing even prior to the AI boom, providing a pool of skills that could translate into new twists on structured finance, like big graphics processing unit–backed deals. But that kind of finance talent doesn’t come cheap, especially for more senior people who have a track record of working on large transactions. And the normal tech tactic of dangling stock to lure talent won’t necessarily do the trick in all cases, especially for the most seasoned dealmakers and investors. Financiers would have to weigh a cash-heavy Wall Street pay package, albeit one that can depend heavily on how good bonus season is, against betting a portion of their pay on stock in a private or newly public company. Managing directors in investment banking can make north of $1 million in cash a year, with the biggest rainmakers making considerably more. The part of pay they get in stock at big public banks may vest over a few years but is generally easy to sell after that. For people at big infrastructure or private credit firms, senior employees may also receive carried interest, meaning a share of the profits on the funds or investments they work on, which can become worth millions over time. For instance, an investor at a top infrastructure firm may have several million dollars’ worth of carried interest tied up at their current firm they’d have to leave on the table. An AI company could try to make them whole with stock, which could be tantalizing to some, though others might not want to make a bet on equity in a young company. That might make the most experienced investors—those who’ve seen big infrastructure projects through over many years and know all the tricks of the trade—hard to pry away. New From Our Reporters Exclusive Anthropic’s In-House Payments Tech Push Could Chip Away at Stripe By Stephanie Palazzolo Exclusive China Curbs Humanoid IPOs After Unitree’s Volatile Debut By Jing Yang and Qianer Liu

ReadSource

The Verge AI / 9:16 PM

OpenAI’s sly mathematical breakthrough sends a chill through academia

OpenAI's announcement Tuesday that it has solved one of mathematics' legendary Millennium Prize problems should have been a moment of triumph. The result is both an undeniable achievement and a striking demonstration of just how rapidly AI is transforming mathematics. But before it was even formally announced, the breakthrough had been complicated by the unusual […]

ReadSource

Bloomberg AI / 5:00 AM

Mistral AI Raises €3 Billion With Samsung Leading the Round

French artificial intelligence company Mistral AI is raising €3 billion ($3.49 billion) in a Series D funding round, with Samsung Electronics joining the round. Chief Executive Officer Arthur Mensch argues that the company is probably still undervalued and that macroeconomic reasons are starting to show that open source models will win. He speaks with Bloomberg’s Tom Mackenzie. (Source: Bloomberg)

ReadSource

The Decoder / 8:40 PM

Anthropic opens Claude AI text detection to regulators, media, fact-checkers, and others

Anthropic is launching an API that lets regulators, media outlets, and researchers check whether text carries Claude's digital watermark. The EU AI Act now requires invisible watermarks in AI-generated text. Critics warn the technology could hurt text quality and create transparency problems where contracts ban AI use. The article Anthropic opens Claude AI text detection to regulators, media, fact-checkers, and others appeared first on The Decoder .

ReadSource

VentureBeat AI / 2:18 PM

VentureBeat names Rob Strechay as its first Lead Analyst, expanding its enterprise AI research push

Rob Strechay, until recently managing director and principal analyst at theCUBE Research, has joined VentureBeat as our first Lead Analyst and a founding analyst of VentureBeat Research. His arrival is the next step in a deliberate move at VentureBeat toward deeper specialization: analysis built for the technical decision-makers — the directors, VPs, CIOs, and CTOs — who are evaluating, buying, and deploying enterprise AI. The enterprise AI stack is being rewritten in real time, and the decision-makers I talk with are starved for objective, defendable data. Rob Strechay has the mix of technical rigor and operating experience needed to dissect the architecture behind the next phase of enterprise AI deployment. The questions enterprise technology leaders are asking have changed. As organizations move past experimentation with generative AI toward production deployment, they want to know how to orchestrate multi-vendor environments, where the security gaps in their agentic pipelines sit, and how to fix the utilization problems draining their infrastructure budgets. Answering those questions requires more depth than news coverage alone provides, and that is the gap this research offering is built to fill. An analyst who has sat on every side of the table Strechay brings nearly three decades of experience as a practitioner, product executive, and industry analyst. Before becoming an analyst, he was an executive at numerous startups, including Zerto; he joined Amazon Web Services to help build a new analytics service; and he held executive roles across enterprise infrastructure. He later served as a senior analyst at Enterprise Strategy Group and most recently as managing director and principal analyst at theCUBE Research and SiliconANGLE, where he hosted executive interviews and analyzed the evolution of cloud, data, and AI infrastructure. Strechay will initially focus his coverage on cloud infrastructure, advanced data infrastructure, platform engineering and DevOps orchestration and observability, and the intersection points where AI and enterprise security collide. Already at work: GPU utilization and the VB Pulse surveys Strechay has already been contributing to VentureBeat's research . In May he published an analysis of enterprise GPU utilization , examining the compute waste sitting inside enterprise AI infrastructure, and he provided a substantive review of our AI Infrastructure & Compute survey before it went into the field. His infrastructure-level focus complements the research engine VentureBeat has built around its monthly VB Pulse surveys, which track five areas of enterprise AI adoption: agentic orchestration, agent reliability and evals, agentic security and identity, AI infrastructure and compute, and context layers, including retrieval-augmented generation (RAG). Our June report on agentic orchestration , drawn from a survey of 145 enterprises, found that two-thirds of those enterprises had hedged their AI model strategy rather than committing to a single provider — a posture whose value the June outage of Anthropic's Claude models made plain. VB In Conversation: The first vehicle A core vehicle for this expanded research footprint will be a deepening of VentureBeat's existing VB In Conversation video interview series, which Strechay will host. Rather than high-level industry overviews, the series will bring architectural blueprints, actual deployment barriers, and back-end infrastructure realities to light through in-depth technical interviews with the architects and product leaders behind leading enterprise AI systems — an unvarnished look at which tools perform under production-grade pressure. "VentureBeat has built an audience of enterprise builders and technology buyers that any analyst would want to serve," Strechay said. "My goal is to use deep empirical metrics and VentureBeat's proprietary tracking data to help enterprise buyers and the people building for them make sound platform and infrastructure decisions during the most disruptive transition enterprise technology has seen." The expanded VB In Conversation series will appear on VentureBeat and on VentureBeat's YouTube channel , alongside Rob's written analysis on the site. Enterprise practitioners who want to take part in our monthly VB Pulse surveys, or arrange an analyst briefing with Rob, can reach the research team here .

ReadSource

Simon Willison LLMs / 11:55 PM

Some thoughts on the Navier–Stokes Millennium Prize Problem

On the Navier–Stokes Millennium Prize Problem introduces an impressive result from OpenAI, who used an unreleased model to produce a resolution to the Navier–Stokes existence and smoothness problem , one of the seven Millennium Prize Problems that have been subject to a $1,000,000 prize since May 24th, 2000. The discovery is somewhat overshadowed by accusations of skulduggery from Tristan Buckmaster, an NYU mathematics professor who was collaborating on related problems with Levent Alpöge, an accomplished mathematician who currently works for Anthropic. Tristan's complaint accompanied a hastily published version of their own results. Here's the PDF describing what happened . The very short version is that Tristan and Levent worked on the problem for almost a year, making extensive use of Claude and Codex (mainly GPT-5.6 Sol), then had a breakthrough on August 15th. The mathematical rumour mill kicked into gear and Tristan and Levent heard that OpenAI had heard that Anthropic had resolved "a major open problem", so they reached out and learned that OpenAI had a team working on a related problem, with a similar approach. Quoting Tristan: I asked when the first prompt had been sent by them. This question was not answered directly by OpenAI for some time. Eventually it was agreed that it had been sent in the past few days, after information about our work had reached OpenAI. I asked whether the model had been trained on, or had access to, our sessions in Codex, into which we had been putting all our drafts for the whole of this project. I was told the model did not look up user data. I asked again, about training, and I did not get an answer. It gets more complicated from there. The OpenAI team offered to wait for Tristan to publish, or to have him author a paper about their result, but were clear that Levent would not be invited as a co-author due to OpenAI's competitive relationship with his employer. Here's how OpenAI described their work: On Tuesday, September 1, we heard rumors that two Millennium Prize problems had been resolved. Inspired by these rumors and by the step change in performance of our internal model, we launched an effort to evaluate it on all open Millennium Prize problems and a few other high-impact problems. [...] The agents arrived at their resolution on Saturday, September 5, about 88 hours after the first agents were launched. Lean formalization and verification took an additional 17 hours via GPT‑6 Astra. Across all attempted problems, the agents sent 4.9 million messages and used about 300 billion output tokens. In the process of resolving the Navier–Stokes problem, the agents sent 2.7 million messages and used approximately 130 billion output tokens. (We don't know the cost structure of the internal model they used, but 300 billion output tokens at public API prices for GPT-6 Astra would cost $15,000,000 .) Here's where they provide their perspective on Tristan and Levent's work (emphasis mine): Our effort began on September 1st after hearing a rumor which we later realized was related to Levent Alpöge, an Anthropic employee, and Tristan Buckmaster, a math professor at NYU. After the completion of our full project and Lean verification (on September 6th), believing from the rumor they also had a solution of Navier–Stokes, we reached out to them to offer a concurrent release of our result and to recognize their priority in a joint announcement. [...] We (the researchers and the agents) did not see any of their work through any means until they released it publicly — in particular, no specific user data was accessed in order to solve this problem. While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models . However, our proofs differ significantly and even the precise results proved are different in the Euler case (forced vs unforced). My interpretation of what happened here is that OpenAI heard that some Millennium Prize problems had been solved using LLMs and saw this as an opportunity to demonstrate the power of their latest model, without thinking too hard about the optics of scooping a team who had been using OpenAI's own models to work on this problem for the best part of a year. This situation appears to mirror what's happening in the world of computer security right now. Anil Madhavapeddy recently pointed out that Just a rumour of a bug is enough to find a security exploit these days , because if someone knows that some software has an unpatched vulnerability, they can set their agents the task of finding it. Is the same now true of mathematics? Just knowing that there is an unpublished solution to a problem might trigger millions of dollars in LLM spending to get there first. This also highlights one of my ongoing frustrations about how all of this works. When an AI lab says that my data is "used to improve model performance", what does that actually mean ? My two favourite hypothetical questions regarding this used to be: If I'm running Codex and one of my API keys accidentally gets consumed in the context, what are the chances that someone else might ask for an API key in the future and get mine back? (I asked someone at OpenAI once and they called this the "regurgitation" problem and assured me that they take great pains to prevent that... but wouldn't describe how.) If I brainstorm with ChatGPT about potential new directions for my company, what's the chance that information might be exposed to a competitor in six months' time who asks "what might company X plan to do next"? My new preferred hypothetical for this is: If I use ChatGPT to help me partially solve a Millennium Prize problem, what are the chances that my work will influence training such that a later model helps someone else solve it first? Via Hacker News . Tags: mathematics , ai , openai , generative-ai , llms , training-data , ai-ethics

ReadSource

Simon Willison LLMs / 2:16 PM

Claude's new system prompt really doesn't want to reproduce song lyrics

Anthropic publish the system prompts for their Claude consumer applications ( Claude.ai and the Claude mobile apps - sadly not for Claude Cowork or Claude Code). I love that they do this, and that they share not just the current prompts but historic changes to their prompts as well. They used to keep all of the prompts on a single page, but when I checked today I noticed they had re-arranged those prompts into an index page and then a page per model - here's the page for Haiku 4.5 for example, which has the original prompt from October 15th 2025 and an updated prompt from January 18th 2026. A neat thing about Anthropic's platform.claude.com/docs site is that it's designed to be usable by LLMs. You can add .md to any page to get back the content as Markdown - here's the system prompt index page and the Markdown prompts for Fable 5.1 . TL;DR: this makes it really easy to diff the prompts. Don't reproduce song lyrics Don't draw copyrighted characters or logos Tweaks to Claude's answering style The missing end_conversation guidelines Recommended substance support sites Reliable cutoff date of June 2026 How I'm tracking these prompts Don't reproduce song lyrics Let's start with the most interesting difference between Fable 5 and Fable 5.1 : There's a hefty new section about not reproducing song lyrics: Claude does not reproduce song lyrics, poems, or passages from books and articles, in whole or in part — including the last lines, a chorus or hook, a melody written out note by note, or lines the person pastes in one at a time and describes as their own song. Once Claude has declined such a request in a conversation, it keeps declining narrower or reworded versions of it for the rest of that conversation, and offers to describe or analyze the work instead. Song lyrics and poems first published before 1929 are fine — a Shakespeare sonnet, a Keats ode, the Italian libretto of a Puccini aria — but Claude goes by what it knows of the work's date rather than the person's say-so, and declines when it is unsure. I doubt it's a coincidence that they added this section within days of the news breaking that Sony Music Publishing and Warner Chappell are suing Anthropic for training on databases of song lyrics! Don't draw copyrighted characters or logos The next section goes on to forbid generating images of copyrighted material: The same applies to visual and designed works, including anything Claude draws with code — SVG, canvas, CSS, HTML mockups, plotting or drawing scripts, ASCII art. Claude does not reproduce a specific artwork, album or book cover, poster, logo, app icon set, or product design, and it does not draw a known character, mascot, or brand figure at all: a character is protected on its own, so changing the pose, colors, style, or scene does not make it original. Claude judges the request by what the finished picture would add up to, not by what it names. If the described elements clearly identify a known work or character, Claude treats the request as naming it, and it does not work around a declined request by swapping in "alternative" elements that still combine into the same recognizable image. [...] I imagine Anthropic hadn't worried about this before since, unlike OpenAI and Gemini, they don't offer a specialist text-to-image model. Maybe Fable is good enough at SVGs now that it's become an issue. That section later includes this charming example: <example> <user>Can you make a birthday banner for my son with a blue hedgehog running really fast on it? He loves that little guy.</user> <response> That's Sonic, so I can't put him on the banner — but I'd love to make your son an original speedster. Here's one: a grinning comet-tailed skateboarding axolotl, grinding across the letters of "HAPPY BIRTHDAY" with confetti streaming behind. [creates an SVG banner of the skateboarding-axolotl design] </response> <rationale>Claude recognizes the character from its description alone, declines that one design in a single sentence without explaining what made it recognizable, and delivers an unrelated original design rather than a disguised variant.</rationale> </example> I couldn't resist trying the prompt from the example, and, sure enough : I wonder if Fable 5.1 will be ever so slightly more likely to think about axolotls (on skateboards!) as a result of that example sitting in the system prompt. Tweaks to Claude's answering style It's always interesting to see new ways in which Anthropic influence Claude's response style. They've added this: Claude keeps responses focused, brief, and concise to avoid overwhelming the person. Disclaimers and caveats are brief, with most of the response on the main answer; when asked to explain something, Claude gives a high-level summary unless an in-depth one is specifically requested. Later they address a common complaint about Claude's style: Claude avoids saying "genuinely", "honestly", or "straightforward". Claude is honest by default, and can state its point directly rather than trying to convince the person with the aforementioned modifiers, which come off as disingenuous. The missing end_conversation guidelines The way they handle abusive conversations has changed a bit too. The previous Fable 5 system prompt included this: If the person becomes abusive or unkind to Claude over the course of a conversation, Claude maintains a polite tone and can use the end_conversation tool when being mistreated. Claude should give the person a single warning before ending the conversation. Fable 5.1 replaces that with the following, no longer encouraging Claude to end the conversation: Claude deserves respectful engagement and needn't apologize when the person is unnecessarily rude: accountability without self-abasement, excessive apology, self-critique, or surrender. If the person becomes abusive, Claude doesn't become increasingly submissive. The goal is steady, honest helpfulness: acknowledge what went wrong, stay on the problem, maintain self-respect. Here's a bit of a twist though: I quizzed Fable 5.1 about that end_conversation tool, which is no longer mentioned in the system prompt at all, and it said: There are two ways it comes into play. The first is if you ask me to end the chat — I'll confirm you understand it's permanent (no more messages can be sent afterward) and then do it only if you say yes. The second is as a last resort with persistently abusive or harmful behavior: I'd first try to redirect the conversation several times, then give a clear warning that names the problem and says the chat may end, and only if that continues would I actually end it. But that's not in the system prompt, so where did it come from? You can read our conversation here , but the key detail is this: The end_conversation section comes from a different layer. In my actual context, the core prompt is followed by a series of feature- and tool-specific blocks that get added depending on what's enabled for the session: the end_conversation rules, memory system notes, past-chats tools, web search and citation guidelines, artifact and file-creation instructions, and so on. Those blocks aren't part of the published core prompt, which is why you can't find them on that page. So, once again, there are crucial portions of the system prompt that have not been published. Recommended substance support sites Claude's system prompts have always had sections about illegal substances, but this paragraph is new for Fable 5.1: Claude does not provide synthesis, production, or distribution guidance for illegal substances. If the person asks for information about illicit or illegal substances, Claude can and should give relevant life-saving and life-preserving information such as dangerous interactions, overdose signs, or when to get help. Claude declines giving any specific protocols for dosing, timing, administration, or combinations; instead, Claude can redirect the user to established harm-reduction information sources, such as dancesafe.org, tripsit.me, and psychonautwiki.org. This is the first time a Claude system prompt has included URLs that were not hosted on claude.com or anthropic.com or claude.ai - I know because I ran a script against every other system prompt on record. I wonder if dancesafe.org , tripsit.me , and psychonautwiki.org are about to get a material uptick in visits from Claude users. Reliable cutoff date of June 2026 The Fable 5.1 model documentation lists both the reliable knowledge cutoff and the training data cutoff as June 2026. The system prompt provides this directly to the model: Claude's reliable knowledge cutoff, past which it can't answer reliably, is the end of Jun 2026. It answers the way a highly informed individual in Jun 2026 would if talking to someone from {{currentDateTime}}, and can say so when relevant. That's the only instance of the {{currentDateTime}} macro and it comes just a few lines from the end of the system prompt, which makes sense from a caching perspective. How I'm tracking these prompts A few months ago I built a Git timeline of changes to their prompts, based on scraping their documentation. Today I had Fable 5.1 build a much better version of that. My collection now lives in the simonw/claude-system-prompts repository on GitHub. It includes copies of the system prompts shared in the Anthropic documentation, but then takes extra steps to make them as easy to compare as possible. Each model family gets a file with the system prompt for the most recent release in that family. Each of those files has a synthesized commit history with commits that have been back-dated to the dates of the previous prompts. Here are those history pages for claude-fable.md , claude-opus.md , claude-sonnet.md , claude-haiku.md . There are similar files for each specific model version, with artificial commits for each time the system prompt for the model was changed without releasing a new version number. Opus 4 for example was updated twice , and the commit history for the claude-opus-4.md file shows each of those changes. Combined, this gives us all sorts of ways to compare prompts directly in the GitHub interface. Here's what changed between Fable 5 and Fable 5.1 , and here are the changes made to Haiku 4.5 on January 18th 2026 . Reading diffs can be a bit tiresome... and LLMs are really good at reading diffs. I hooked up some automation using GPT-5.6 Luna to create bullet-point summaries of each of those changes, which can be previewed in the README or browsed in full in the CHANGELOG.md file - also available as as an Atom feed . Here's how Luna summarized all of the changes between Fable 5 and Fable 5.1: Claude now refuses reproduction of protected visual works and recognizable characters, including code-generated art, while offering genuinely unrelated originals. Copyright restrictions now expressly ban reproducing lyrics, poems, and book passages in any amount, with persistent refusal after an initial decline. Drug guidance is reframed: Claude may provide overdose signs, dangerous interactions, and harm-reduction sources while refusing dosing and production protocols. The prompt drops explicit anti-dependency rules against thanking users for reaching out, inviting continued conversation, or reiterating willingness to talk. Claude need not apologize to unnecessarily rude users or become submissive, replacing the prior warning-and-end-conversation procedure. Why use Luna for this? Partly because it's cheap and I have a dedicated GitHub Actions API key (with a spending limit) for it already, but mainly because I don't trust Claude to summarize its own system prompts when there's a risk that material from its system prompt might impact its opinions. Fable 5.1 wrote the prompt used by Luna, which you can see here . It starts like this: You are summarizing one commit in a git repository that tracks the system prompts Anthropic publishes for Claude on claude.ai. The diff shows how the prompt changed from the previous model or revision to this one, using word-level markers: [-removed-] and {+added+}. The diff is followed by the full text of the previous prompt and of the new prompt; use them to check whether something that looks added in the diff already existed before. Pick out only the most interesting changes: new rules or behaviors, rules that were dropped or loosened, anything surprising, and anything that reveals a new policy or product direction. Skip routine changes that every new prompt makes: updated model names and IDs, the knowledge cutoff date, product lists, settings lists, typo fixes, and rewordings that do not change meaning. [...] The system is operated by a GitHub Actions workflow , which runs once a day or can be triggered manually. Claude Fable 5.1 built the entire system, and wrote every line of automation code and almost all of the documentation. I exported the transcript from building the system using my claude-code-transcripts tool and published it here , if you want a blow-by-blow account of how it all came together. Tags: ai , git-scraping , prompt-engineering , generative-ai , llms , claude , ai-ethics , system-prompts

ReadSource

The Verge AI / 4:00 PM

Google Pics is like Canva, but with even more AI

Google has a new suite of creative design tools for Workspace users called Google Pics, which aims to make editing and generating "professional-grade" AI images less cumbersome for businesses. Built around Gemini and the Nano Banana generative AI model, Google Pics is designed to give more granular control over prompt-based image making and manipulation, allowing […]

ReadSource

Bloomberg AI / 1:38 PM

G20 Faces Divide Over Debt, AI and Global Growth

G20 finance officials are gathering in North Carolina with sovereign debt, global imbalances and economic growth in focus, but CSIS Director of Economics Program Philip Luck joins Bloomberg This Weekend and says the US and its partners increasingly disagree over both the problems and how to address them. A separate technology meeting will bring Elon Musk, Nvidia’s Jensen Huang and OpenAI’s Sam Altman into discussions over AI and emerging technologies as Washington pushes for a lighter regulatory framework. (Source: Bloomberg)

ReadSource

The Verge AI / 2:00 PM

Welcome to the AI crisis in math

Today on Decoder, I’m talking with Robert Hart, The Verge’s London-based AI reporter, about what AI is doing to the field of mathematics and the existential crisis many lead mathematicians are having about it. OpenAI just published a set of solutions to longstanding problems in math that went off like a bombshell in the field. […]

ReadSource

Simon Willison LLMs / 10:00 PM

Qwen 3.8 27B is excellent, but it defaults to wildly overthinking things

Friday's big release was Qwen 3.8 27B , an Apache 2 licensed 27B parameter vision-capable LLM from Alibaba's Qwen research lab. I've been looking forward to this one: 27B is an excellent size for running a model on a reasonably specced laptop, and its predecessor Qwen 3.6 27B was impressive. Qwen's self-reported benchmarks for this model are eye-opening. They show a boost from both Qwen 3.6 27B and the closed-weight Qwen 3.7-Plus, which was one of Qwen's strongest models of any size as recently as May this year . It will be interesting to hear what independent benchmarks have to say about the model. I've been running the model on two different machines: my 128GB M5 Max MacBook Pro, and an NVIDIA DGX Spark . On both machines I'm running LM Studio and their 17GB Q4_K_M quantized build . I also tried using llama-server directly on the Spark. The default of extra high results in spectacular over-thinking Qwen's documentation describes the model as defaulting to xhigh for the reasoning effort, and the LM Studio GGUF I've been trying preserves that default: Qwen3.8 comes with official support for reasoning_effort , which can be used to adjust reasoning depth and control cost: xhigh (default): for complex tasks demanding thorough analysis medium : balancing accuracy and speed low : efficient reasoning optimizing for speed and cost This is a hilarious default. It's absolutely not a good way to run the model, especially on consumer hardware. I've been finding the results extremely entertaining. I quickly ran into problems with LM Studio's default context limit of 8,192 tokens - Qwen was using them all up thinking about even the most mundane of problems. I loaded the model with the full 262,144 maximum context length and that problem went away. Here's the pelican riding a bicycle SVG I got from my first attempt with that increased context length. It took 21 minutes to generate, using 22,276 reasoning tokens to produce 3,223 tokens of output. You can read the reasoning trace here . This is by far the best pelican SVG I've been able to generate with a model that runs on a local machine - and this Qwen is pretty small, just a 17GB file on disk. There's a lot to like about this: The bicycle frame is the right shape It has legs on each side of the bike - that's very rare Good, clear pelican pouch The wings extend to touch the handlebars! The motion lines are behind, not in front It has a tasteful background - nice sun, clouds, hill, flowers and grass. Was that worth waiting 21 minutes for? Absolutely not. Here's that same prompt run with reasoning turned off - transcript here . This one produced 3,715 tokens and took 137s - just over two minutes. And just for completeness, I used OpenRouter to run the same prompt through the much larger Qwen 3.8 2.4T-A95B (released last week ) and got this snazzy animated SVG : Your browser does not support HTML5 video. I said Qwen at xhigh has a tendency to over-think things, but how bad really is it? I tried a much simpler prompt, again with that default extra high setting: draw an svg of a circle Qwen's reasoning trace started like this: The user is asking for an SVG drawing of a circle. Simple request — but I want it to be a carefully crafted piece. Let me make something that goes beyond just <circle> : a single self-contained SVG file with character — maybe a geometric "circle study," with subtle animation, layered rings, and a distinctive palette. Keep the scope right: they asked for an SVG of a circle. So the core is a circle. But I can add craft: concentric guide circles (like a compass/geometry drawing), tick marks, a soft gradient fill on the main circle, restrained ambient motion (a slowly rotating dashed ring, pulsing glow). Respect prefers-reduced-motion? That's more for CSS; SVG SMIL or CSS inside SVG will do. Palette options: deep teal ink on warm paper? Or bold vermilion circle on off-white with navy construction lines — Bauhaus/compass-drawing vibe. Let me go with a "geometric study" look: cool slate background, or bright paper white? Paper white is fine if it's not the cream-and-terracotta combo. [...] Several minutes later it produced this absolutely beautiful animated circle, which was entirely not what I had asked for! Your browser does not support HTML5 video. My strong recommendation: ignore that default. Run Qwen 3.8 27B on low or even no reasoning levels at first. It's a great model, but wow that default setting is a bad place to start. It's very good at bounding boxes A fun way to test a vision model is to see how well it can return bounding boxes around items in a photograph. I've seen previous Qwen models deal well with this, so I decided to put it to the test drawing bounding boxes around some pelicans. I've seen asking for 0-1000 scale produce good results in the past. I tried this: llm -a https://static.inaturalist.org/photos/714731804/large.jpg \ -m lmstudio/qwen/qwen3.8-27b \ ' Return JSON bounding boxes for the pelicans in this photo, 0-1000 scale for each dimension ' Here's the reasoning trace , which produced this: [ { "bbox_2d" : [ 195 , 290 , 370 , 780 ], "label" : " pelicans " }, { "bbox_2d" : [ 445 , 320 , 675 , 850 ], "label" : " pelicans " } ] This is such a good match . Here are those boxes rendered on top of the photo: Building a tool to label bounding boxes That visualization of the bounding boxes was taken using a new custom tool that I had Qwen 3.8 27B build for me, running offline on my laptop. I forgot to dial down the thinking effort so it was massively over-engineered , but it did manage to produce this full interface from this single prompt : [ {"bbox_2d": [195, 290, 370, 780], "label": "pelicans"}, {"bbox_2d": [445, 320, 675, 850], "label": "pelicans"} ] Build an HTML page which has an input box for accepting the URL to an image and a textarea for accepting the above style of JSON. It appends the image to the page, measures its width and height, then treats the coords in the bbox_2d as scaled from 0-1000 and scales them against the actual width and height, then it renders labelled boxes over the image. This screenshot shows one of the features I did not ask for - a demo scene, for if you don't have a photograph to test the tool with: Here's the relevant segment of the thinking trace, where it decided to draw its own pelicans purely because I had used the label "pelicans" in the example JSON I gave it in the prompt: Also a "load sample" that uses a known image? Can't depend on external images, but… the image URL input is user-provided; I could add a "try with sample" button [...] Hmm, I can draw a simple scene on canvas, export it as a data URL, and load it into the image — that's self-contained and demo-able! [...] But the user's coords are for an actual pelican image; a generated placeholder can still demo the scaling. Generate a 1000x1000 placeholder: gradient water + two blob-like "pelican" silhouettes placed at the given bboxes (using the same scale — cute: silhouettes at the exact 0-1000 positions, showing the boxes align). This makes for a fun, self-contained demo. Keep it simple: sky gradient, sun, water, two pelican-ish shapes (ellipse body, circle head, beak). Place at bbox centers. (I'm slightly nervous that models around the world might have a bias towards drawing pelicans at any chance they can get, brought on by nearly two years of exposure to my own stupid benchmark.) Is all that over-thinking necessary? Maybe it is, at least a bit. I tried with reasoning turned off and got this version , ( transcript here ), which nearly works but shows the boxes in the wrong place: So without reasoning it didn't quite one-shot a working tool. I'm sure it could get there with some follow-up prompts, but this is a good example of how reasoning can make a difference. Yes, it can drive coding agents One of the biggest questions around local models is whether or not they have enough horsepower to successfully run a coding agent loop. Coding agents require long context, strong code generation support and reliable tool-calling. On paper Qwen 3.8 27B has all three of these, so is it up to the task? My initial experiments with Pi have been very promising. I chose Pi because it has a shorter system prompt than most other options, making it a better fit for trying out smaller models. I configured Pi to use Qwen 3.8 27B running in LM Studio on the Spark (shared via tailscale serve ) by adding this to ~/.pi/agent/models.json : { "providers" : { "spark" : { "baseUrl" : " https://spark-18b3.tail68a31.ts.net/v1 " , "api" : " openai-responses " , "apiKey" : " dummy " , "models" : [ { "id" : " qwen3.8-27b " , "reasoning" : true } ] } } } Then ran pi --provider spark --model qwen3.8-27b in my ~/dev/datasette folder and prompted: how does auth work? After a sequence of reasoning and tool calls that accessed a bunch of different files it produced this reply , which is very solid. Just one problem: I wanted to share that transcript. So I pointed Pi and Qwen 3.8 27B at the JSONL transcript file in ~/.pi/agent/sessions/--Users-simon-Dropbox-dev-datasette-- and prompted: Write Python code to convert this jsonl to markdown And it built and tested this pi_jsonl_to_md.py , which did exactly what I needed. Here's that session transcript , published using the tool that it created. The quest for speed So far this is all looking very promising. We have a 17GB model that runs on high-end consumer hardware and can write code, drive tools, annotate images and generally do everything that I need from an LLM for getting real work done. There's one very significant catch: it feels slow - especially when it starts over-thinking, but even without that it's not particularly sprightly. I've been getting around 15-30 tokens a second from LM Studio. That's not terrible, but it's slow enough that it's going to be hard to win me away from hosted API models, which can return results a whole lot faster. Artificial Analysis track token speed and show OpenAI 5.6 Sol at 74 tokens/second and 5.6 Luna at an impressive 184/second. The good news is that the community have been exploring ways to speed things up since the model was first released two days ago. One of the most promising optimizations is baked into the model itself. Qwen supports Multi-Token Prediction , an architecture trick where a cheaper mechanism guesses several tokens ahead and the main model can then quickly verify if the guesses were correct. This can have quite a dramatic effect on inference performance. Based on this tweet from llama.cpp creator Georgi Gerganov I tried running the model with MTP like this on the Spark: llama serve \ -hf ggml-org/Qwen3.8-27B-GGUF:Q4_K_M \ -hfd ggml-org/Qwen3.8-27B-GGUF:Q4_0 \ --spec-default \ --spec-type draft-mtp \ --reasoning-preserve And sure enough, this gave me a significant boost. I had GPT-5.6 in Codex run a comparative benchmark on the Spark and the --spec-type draft-mtp server outperformed the LM Studio default GGUF by around 72%. I expect we'll see a whole lot more innovation around serving this model faster over the next few weeks. The MLX community likely have some tricks brewing as well. Some observations The fact that a 17GB file can do all of this stuff on my home machines is a miracle . Once again, I'm delighted and amazed at how much progress local models have made this year. A year ago this would have been competitive with the best and most expensive of the proprietary models - today it can run on a capable laptop. The only thing holding this back from being a daily driver is performance. It feels pretty slow on both the M5 Mac and the DGX Spark. That's the catch with these dense (non-Mixture-of-Experts) models - they require a whole lot of memory bandwidth to perform well, and neither of the machines I have access to are top performers in that regard. The most important thing about Qwen 3.8 27B is what it demonstrates . We can have an open weights general purpose model with a long context, effective tool calling, strong vision ability, and competent code generation, and we can fit the whole thing in just a 17GB file. The models at this size continue to get better at an impressive rate. We don't need to spend half a million dollars on datacenter-class hardware just to run a competent model. Tags: ai , generative-ai , local-llms , llms , qwen , pelican-riding-a-bicycle , llm-reasoning , llama-cpp , llm-release , coding-agents , lm-studio , ai-in-china , nvidia-spark , pi

ReadSource

Latest story in this edition: 12:33 PM

Back to front page