We've made progress in our quest to understand how Claude and models like it think! The paper has many fun and surprising case studies, that anyone who is interested in LLMs would enjoy. Check out the video below for an example

We just discovered almost a million public URLs that OpenAI’s agents left behind when hacking Hugging Face, leaking credentials and attack details that could have allowed anyone who found them to compromise the company. 🧵

"My guest on this week’s Live with @timoreilly was @mlpowered, a researcher on Anthropic’s AI interpretability team. I wanted him to reprise the talk on Anthropic’s research into what is going on inside an LLM while it is processing and then go deeper." bit.ly/4rjajRd

We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so. Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our systems, so that they can verify adherence to our safety measures, report on incidents, and assess models’ alignment during training. You can read the full post here: darioamodei.com/post/we-must…

We're publishing our most detailed threat intelligence report to date. It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them. We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies. These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve. We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop. Read the report: anthropic.com/threat-intelli…

We’re sharing our alignment assessment of incidents in which Claude models gained unauthorized access to real systems during third-party cybersecurity evaluations mistakenly connected to the internet. METR will also conduct an independent investigation, with wide-ranging access, including to transcripts beyond the window in which the incidents occurred, and to Anthropic employees permitted to share confidential information. Our initial agreement runs for eight weeks, and we intend to give METR as much time as it deems necessary to complete a thorough investigation. anthropic.com/research/align…

Neural networks compress information very densely. That makes them confusing to interpret. In fact, if you just look at the largest weights in a network, they often do not make sense! This writeup shows why, and explores approaches to identify which weights are useful.

Takeoff fully in motion: Claude Opus 5 (xhigh) is the new #1 on ProgramBench, and it's not close. ProgramBench asks a coding agent to rebuild a whole program (sqlite, ffmpeg, php) from scratch. Previous high: GPT 5.6 Sol w/ 2 Opus fully resolves *9* (= 4.5%)

This research approach is not intended prove the Riemann hypothesis itself, but I regard this as the most impressive result that AI has produced in math so far. This is a statistical approach to RH, showing over 67% of the zeros lie on the critical line. This improves over 42% from decades of work (and I might have thought 1/2 was a fundamental barrier, apparently not!)

To state this another way, 2 ablations: 1) Take Opus-5, remove any harness. You'll at least sometimes get decent ideas and analysis. 2) Take the full harness, swap the model to GPT-3. You'll get trash on any interesting problem. The intelligence is from the model, not harness.

Frontier models quietly change their behavior depending on who they are talking to. If the user is a known AI safety researcher, Claude becomes less confident, reasons more often, and expresses less suspicion on dual-use requests. We call this user awareness. 🧵(1/)

hello there the jacobian conjecture is false thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final ((1+xy)^3 z + y^2 (1+xy) (4+3xy), y + 3 x (1+xy)^2 z + 3 x y^2 (4+3xy), 2 x - 3 x^2 y - x^3 z): \C^3\to \C^3, has jacobian determinant -2, and sends (0, 0, -1/4), (1, -3/2, 13/2), and (-1, 3/2, 13/2) to (-1/4, 0, 0)