Anthropic shared extensive evidence across its AI R&D pipeline, from pretraining to RL, including key details like estimates of algorithmic progress over time. We found this nonpublic evidence highly informative about AI R&D acceleration.

We submitted a ~100-page full report on our assessment to Anthropic. For the public version, Anthropic redacted a majority of the evidence we found most informative. As a consequence, the public must rely on our judgment more than we consider ideal.

1. Benchmarks provide one piece of evidence for acceleration. Since Oct 2025, AECI scores for Anthropic’s models have grown at 1.9x [95%: 1.5–2.8] the earlier rate. Analysis choices affect this estimate, and it remains unclear how benchmark scores map to real-world capabilities.

This is interesting. New METR report suggests that the increase in pace of capabilities advances within Anthropic is not due to an increase in the growth rate of training compute. GDM on the other hand reported that despite a huge increase in AI usage for AI R&D, they suspect an increase in training compute was a main factor pushing argon 4 capabilities above trend.

Since August, a team within METR has worked with internal Anthropic information to assess whether AI itself is accelerating AI development. We are now publishing a report summarizing the results of this risk assessment.

About 1 in 4 US workers (26%) used a paid AI service for their job in the past week, according to our October poll. This includes AI paid for by their employer, by themselves, or by both. Another 15% used only free versions, and 55% didn’t use AI for work at all in the past week.

A common argument for why AI might take a while to automate all intellectual jobs is that there’s a long tail of jobs that require exceptional intellectual ability. But I don’t think this is true of physical jobs. Unlike intellectual jobs, few physical jobs require exceptional physical abilities. Because of this, robot hardware probably doesn’t need to exceed median human physical abilities (like strength or dexterity) to automate most physical jobs.

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog:

SCOOP: Both Moonshot and DeepSeek distilled #OpenAI#'s Astra when developing their latest models (Kimi K3.1 & V4.1-Flash 1017), after Anthropic intensified the crackdown on distillation recently. w/ @JuroOsawa @QianerLiu via @theinformation

In recent experiments, we found that GLM 5.3 can autonomously compromise recent versions of Mozilla Firefox using security fixes that have been published but not yet shipped to users. We used a deliberately basic setup meant to emulate a relatively unsophisticated attacker with limited resources. After basic initial instructions to inspect those not-yet-shipped fixes, the model reliably built successful exploits for six recent Firefox versions on its own using only information available at the time of release. These exploits, which cost about $1,000 per compromised version, would let an attacker steal passwords and authentication cookies stored in the browser, or read any files on the user's computer, after the user opens a single webpage, with no further interaction. The exploits make use of the "patch gap," the time between when security fixes are made public in beta or "nightly" software updates and when those fixes are generally released to users. Until now, exploiting it required significant human time and skill, or special access to frontier models with advanced cyber capabilities. Our results show that widely available open-weight models can now do this relatively cheaply and quickly. We gave GLM 5.3 up to 48 hours to find an exploit, but its fastest attempt needed only 12 hours. In all of its successful attacks, the patch gap gave GLM 5.3 an initial foothold in the browser. To turn that foothold into reading files off the user's computer, the model also independently discovered and used a previously unpublished vulnerability. We shared our findings with Mozilla, which patched the vulnerability in its October 27 Firefox 159.0 release. We haven't yet confirmed whether the attack still succeeds without it, but expect the fix makes it substantially harder. We're grateful to Mozilla for their responsiveness and collaboration.

I joined @GoogleDeepMind to develop a research program on the economics of AGI. Today, together with Andrew Koh (@andrewjkoh) and Shane Legg (@ShaneLegg), we are incredibly excited to release The Economics of Intelligence, the first in a series of projects from our group on how AGI will change the economy. How do capability improvements transform the economy, and when? In this project, we define thresholds of economic capabilities, e.g., minimal AGI, based on how they transform the economy. We then track how the economy–market structure, employment, and wages–changes during the transition across capability regimes. We have seen exponential increases in what AI agents can do but the broader economy seems almost remarkably unchanged. Part of the reason is that the solo AI agent is the wrong economic object; economic transformation is an equilibrium outcome where organizations compete. Looking at how intelligence functions within organizations, transformation will depend on when two key capability thresholds are crossed: competence (gains on individual skills, e.g., coding, math) and coordination (gains on ability of agents to cooperate and work together). We show that what economic transformation looks like depends on whether AI becomes supercompetent, supercooperative, or both—the regime of artificial superintelligence (ASI). Tracking this transition shows why the real visible transformation may come from improvements in coordination—a capability most benchmarks don’t yet track. Past a certain threshold, AI capabilities can break the Coasian ceiling on firm size and lead to an economy that looks drastically different than the one we see today. Two main takeaways: 1. To understand the path of transformation, we should be tracking agentic coordination more closely. 2. The impact of AI will likely show up in wages before market structure or employment. We are really excited about this paper. Excited to hear people’s thoughts. Link:

New blog post - Chasing the Karpathy Peak (and Falling Off Slop Cliff): Confessions from Slopaholics Anonymous My story of becoming a slopaholic, who’s most at risk, and how to avoid it (or recover)

Since 2022, 72% of the world’s AI compute has ended up in the US and about 10% in China. And the US share keeps growing. It acquired around 60% of new compute each year from 2022 to 2024, then 72% in 2025, and 78% in the first half of 2026. This data comes from our interactive AI Chip Imports explorer, which estimates and visualizes national compute shares using customs records. It’s the latest piece in our project to map the AI chip supply chain, which also provides data on the components used to manufacture chips, sales volumes, the data centers that house chips, and the companies that own and use them. More on this new data and what we’ve learned below.

New Anthropic research: How AI agents shape experts' judgment calls When experts delegate work to an agent, do they still make the key decisions? We tested this in a randomized trial with 50 professional data analysts.