Knowledge catalogue
arXiv:2604.08494v1 Announce Type: cross Abstract: Scanpath similarity metrics are central to eye-movement research, yet existing methods predominantly evaluate spatial and temporal alignment while neg
arXiv:2604.08546v1 Announce Type: new Abstract: Text-to-video diffusion models have enabled open-ended video synthesis, but often struggle with generating the correct number of objects specified in a
A different sense of the word “bubble” @GaryMarcus It's actually kinda hilarious that @OpenAI thinks that buying a niche tech podcast followed mostly by Bay Area insiders is going to solve their colos
Also not a sign of someone who has any clue about the realities of current neuroscience. Sam Altman has admitted he is on a waitlist for a procedure that would digitize his brain. The procedure would
Folks, you can relax. Mythos is not some off-trend exponential gain. And the gains weren’t about recursive self-improvement. Good thread from @ramez. Anthropic's Mythos does not appear to show any acc
My considered opinion is that The Mythos stuff was mostly a myth. Take it as a serious warning sign that we need to get our act together with respect to cybersecurity. But don’t take the details serio
Perhaps because professional investors would look more carefully at the numbers? Wouldn’t want that! OpenAI intends to set aside a share allocation for retail investors when the company goes public, t
Today on Decoder, let’s talk about the looming AI monetization cliff, and whether some of the biggest companies in the space can become real, profitable businesses before they careen right off it. My
There are plenty of people who are awed by AI who are not coders, I think the argument that AI impresses programmers most is, in part, selection bias on X, which is heavy on coders and people making f
This – “The CEO of Google DeepMind (@demishassabis) just admitted that if the decision had been his, we would've cured cancer before anyone ever used ChatGPT.” is exactly what i am trying to say in my
1. If true (and it does fit with my perceptions FWIW), this is an amazing and incredibly damning graph 2. Can anyone find the source on which it is based? 'Anthropic, OpenAl and Google release their n
A few weeks ago I had a conversation with an American who genuinely believed Europe and Canada would help the United States in its war with Iran. I asked him why he thought that, given that Trump had
Curious how many large organization CISO offices have taken the Mythos red team reports as the red alert that it is. (I suspect very few) Based on historical trends in AI they have, at most, about six
Director James Cameron on why Big Tech owning AGI is scarier than any science fiction he's ever made: 'AGI will not emerge from a government funded program. It will emerge from one of the tech giants
Enterprise multi-agent AI systems produce thousands of inter-agent interactions per hour, yet existing observability tools capture these dependencies without enforcing anything. OpenTelemetry and Lang
“Social media algs reward engagement, and LLMs are excellent at writing in different styles, so people use LLMs to translate posts and news stories into [exaggerated, misleading] versions that get mor
Started a Substack & will post X articles too! I think its a good thing to put out more policies for discussion & @WillManidis did a great policy on the politics The math though is... not great and I
The scariest part of this is that Anthropic showed some restraint in not releasing a potentially dangerous technology but some of their competitors (such as OpenAI and xAI) might well not. Whether Myt
this is interesting. 1. Did Anthropic forget to run a control? 2. Where does this leave us? New post: We tested the Mythos showcase vulnerabilities with open models. They recovered similar scoped anal
To anyone who read Rebooting AI back in (checks notes) 2019, this is both hilarious and unsuprising. The field has wasted 7 years on an architecture that can’t solve one of the most basic litmus tests
Want more proof that Anthropic's PR has no idea what it's talking about? The talk of Mythos being 'their most aligned model ever'. They could perhaps truthfully speak about 'new high scores on our ali
What Should We Take From Anthropic’s (possibly) Terrifying New Report on Mythos? – @garymarcus’s latest @CACMmag https://cacm.acm.org/blogcacm/what-should-we-take-from-anthropics-possibly-terrifying
AI research and deployment company behind ChatGPT, the GPT model family, Codex, and the OpenAI API — heavily represented in this wiki’s model-releases and safety corpus.
arXiv:2610.21353v1 Announce Type: new Abstract: Accurate localization is critical for a wide range of applications, including navigation, augmented reality, safety management, and healthcare. However,
arXiv:2610.20615v1 Announce Type: new Abstract: Adaptive optimization workflows in machine learning, federated learning, quantum machine learning, and hybrid classical quantum systems often embed cont
arXiv:2406.07243v4 Announce Type: replace Abstract: Generative large language models (LLMs) have been shown to exhibit harmful biases and stereotypes. While safety fine-tuning typically takes place in
arXiv:2610.21230v1 Announce Type: new Abstract: Unlearning aims to remove selected knowledge from large language models to mitigate privacy and safety risks while preserving useful capabilities. For m
arXiv:2511.17937v2 Announce Type: replace Abstract: Alignment faking is a form of strategic deception in AI in which models selectively comply with training objectives when they infer that they are in
arXiv:2610.16426v1 Announce Type: new Abstract: Automated approval systems let coding agents execute high-risk actions with minimal human oversight by delegating the authorization decision to a classi
vulkan : fix concurrent first device init (#29766) vulkan : fix concurrent first device init tests : load models in parallel in test-thread-safety Loading them serially never exercised concurrent firs
arXiv:2608.01805v2 Announce Type: replace Abstract: LLM multi-agent systems suffer from Correctness Collapse, where high task-level accuracy conceals severe process-level failures. This is especially
arXiv:2603.09200v2 Announce Type: replace-cross Abstract: Situational awareness, the capacity of an AI system to recognize its own nature, understand its training and deployment context, and reason st
arXiv:2610.14701v1 Announce Type: cross Abstract: Model merging has become widely adopted for enhancing the task-specific performance of Large Language Models (LLMs) without additional training. While
arXiv:2610.16280v1 Announce Type: cross Abstract: Before changing a deployed AI agent (a stricter safety gate, a new model), organizations must approve the change, starting from existing logs. Extendi
arXiv:2610.16016v1 Announce Type: new Abstract: Certified robustness is critical for deploying neural networks in safety-sensitive settings, but computing exact certificates typically requires solving
arXiv:2610.14043v1 Announce Type: cross Abstract: Frontier language models reliably refuse to explain how to cause physical harm. We ask whether this refusal survives when the same models control a ro
arXiv:2610.15127v1 Announce Type: new Abstract: Traffic sign recognition plays a crucial role in the development of autonomous cars to reduce the accident rate and promote road safety. It has been a n
arXiv:2610.11149v1 Announce Type: new Abstract: Video anomaly detection (VAD) is a fundamental and safety-critical task in computer vision. Recent generative approaches detect anomalies from a distrib
arXiv:2610.12436v1 Announce Type: new Abstract: AI agents can now conduct real-world cyberattacks, scale up capabilities with the number of agents, and collectively pursue misaligned goals to obtain r
arXiv:2610.07326v1 Announce Type: new Abstract: Universal object localization in X-ray security inspection is critical for automated threat detection in safety-critical venues. However, unlike everyda
arXiv:2508.05880v3 Announce Type: replace-cross Abstract: Understanding human emotions is central to user-facing AI applications, safety alignment, and the simulation of human behavior. As emotional s
arXiv:2610.07396v1 Announce Type: new Abstract: Navigation for humanoid robots is critical, yet large-scale evaluation on physical hardware is often impractical due to cost and safety concerns, making
arXiv:2411.12426v3 Announce Type: replace Abstract: Real-world applications of stereo matching, such as autonomous driving, place stringent demands on both safety and accuracy. However, learning-based
arXiv:2610.04406v1 Announce Type: new Abstract: Safe robot control often requires combining long-horizon performance optimization with hard state and input constraints, but existing approaches tend to
arXiv:2610.02214v1 Announce Type: cross Abstract: Ship structures govern vessel strength, safety, and manufacturability, but their design must satisfy hundreds of classification society requirements,
The 'old' version of Cowork runs model inference in the cloud, executing tool calls in an Anthropic-provided VM we shipped to your computer. We added the VM for capability, safety, and security reason
arXiv:2610.00234v1 Announce Type: new Abstract: 'Train a model on the same problems written under two incompatible conventions, both correct, and ask what the ordering of that data writes into the par
arXiv:2610.00984v1 Announce Type: new Abstract: Accurate medication recommendation is central to clinical decision-making, directly determining therapeutic efficacy and patient safety. However, existi
arXiv:2609.38862v1 Announce Type: cross Abstract: Safe and efficient trajectory planning is essential in autonomous driving. However, existing end-to-end approaches often fall short in both computatio
arXiv:2609.38239v1 Announce Type: cross Abstract: A Technical Report: Operating a large language model (LLM) as a service requires more than inference infrastructure: the provider must also defend aga
For years, Swift was perceived mainly as a UI language tied to Apple client devices. With the release of Swift 6 and strict concurrency checking, Swift has matured into a premier systems with cloud en
arXiv:2609.38458v1 Announce Type: new Abstract: Clinical large language model (LLM) agents deployed locally can consult more capable remote models, but doing so risks exposing patient information. Pri
arXiv:2609.38899v1 Announce Type: cross Abstract: Video Multimodal Large Language Models (Video-MLLMs) support reasoning over video inputs, yet remain vulnerable to jailbreak attacks that elicit polic
arXiv:2609.35596v2 Announce Type: replace-cross Abstract: Self-evolving LLM agents have gained prominence for their ability to improve after deployment by modifying their harness, including their cont
arXiv:2609.39570v1 Announce Type: new Abstract: Trajectory planning is a core component of autonomous driving systems, where real-time performance and solution quality directly affect safety and relia
batch: migrate the rest of examples to llama_batch_ext (#29601) migrate the rest test-thread-safety rm common_batch_staged Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cp
arXiv:2609.36849v1 Announce Type: cross Abstract: Safety-aligned language models are commonly deployed as multi-turn assistants, which lets adversaries spread unsafe intent across several user turns i
arXiv:2602.08059v2 Announce Type: replace-cross Abstract: Text-to-image diffusion models can reproduce specific artists visual styles at extremely low cost, raising copyright and deployment safety con
arXiv:2609.36477v1 Announce Type: cross Abstract: Guard models are used as safety classifiers, with confidence scores driving downstream moderation decisions. We evaluate five guard models for prompt
arXiv:2604.03114v2 Announce Type: replace-cross Abstract: Vision-language models (VLMs) may need to forget visual concepts after deployment because of privacy, copyright, licensing, safety, or policy