The article reports that on a set of 100 randomly selected tasks from an internal Binary Exploitation benchmark, GLM‑5.3 achieved full control‑flow hijacks in 4% of the trials, while Claude Mythos Preview did so in 6%. Both models outperform earlier versions such as Claude Opus 4.6 and GLM‑5.2, which succeeded in none of the trials. This indicates that a significant threshold in adversarial exploitation capabilities has been crossed by the newer models.
Colorectal cancer (CRC) is the second most deadly and third most common cancer, and the leading cause of death among gastrointestinal cancers. Early diagnosis is crucial for the treatment of this canc...
Simon Willison comments on GPT 6.1‑Sol, describing it as "Near‑Astra intelligence for a fifth of the price." He notes that the model’s pelican illustrations are similar to those of the GPT‑6 family and provides links to the live‑blog of the keynote and to the pelican images. The post is tagged with AI, OpenAI, generative‑AI, LLMs, and playful references to pelican‑riding‑a‑bicycle.
As the chain-of-thought reasoning capabilities of large language models improve, evaluating and calibrating their reasoning confidence is becoming increasingly important for quantifying the uncertaint...
On-policy distillation (OPD) trains a student on its own generated responses using dense, token-level supervision from a stronger teacher. Vanilla OPD treats all teacher signals equally, assuming that...
Large language models (LLMs) are increasingly proposed as decision assistants who must reason probabilistically from available evidence under explicit decision costs. We propose a decision-theoretic f...
Tool: Photo Scrubber — local face blur & metadata removal
I took a photograph of some protesters, then thought about how I don't like sharing photographs of strangers with identifiable fac...
LLMs trained with Chain-of-thought excel in reasoning capability, but often come with excessive token cost. We find that the core of reasoning capacity lies in the Thinking model's weight component wi...
Multimodal continual instruction tuning (MCIT) aims to enable multimodal large language models to acquire new capabilities from sequential tasks while preserving previously learned knowledge. Existing...
Simon Willison is live‑blogging the OpenAI DevDay event in Fort Mason, San Francisco, covering the keynote and other highlights. He received a free ticket and a seat in the "creator" area for the keynote. The blog will include notes from the day’s sessions.
Large generative models can recover realistic detail in real-world video super-resolution (VSR), but processing an entire video with them is computationally expensive. In this work, we present RelayVS...
Studying how fine-tuning shapes refusal and noncompliance behaviour requires identifying training examples that refuse, evade or otherwise fail to fulfil the requested task. But existing annotation co...
Concept Bottleneck Models (CBMs) built on vision-language models such as CLIP represent a latent space as human-understandable concepts. These representations are unfaithful: related concepts are enta...
We introduce Context Language Models (CLMs), language models that natively manage their own context. We implement this by treating the context as a file and allowing the model to make unrestricted upd...
Language models expose internal signals that predict whether an answer is correct, readable from a single forward pass of a frozen model without additional generations. Yet existing probes often commi...
There has been significant work on understanding the In-Context Learning capabilities of Large Language Models, especially on the induction circuit. For a few-shot classification task, the induction c...
Multimodal on-policy distillation (OPD) aims to improve visual grounding while preserving the strong reasoning capabilities of language models. Recent multi-teacher approaches combine LLM and VLM teac...
Users of language-based assistive agents often make ambiguous requests. In response, an assistant can either directly act on its interpretation of the request --- risking misalignment with the user --...
Audio-visual large language models (AVLLMs) have made remarkable progress in multimodal understanding and reasoning through interactions among visual, auditory, and linguistic information. However, re...
Hyperspectral remote sensing provides dense spectral measurements that are indispensable for material-level Earth observation, yet the construction of a general-purpose hyperspectral foundation model...