Knowledge catalogue
arXiv:2604.08987v1 Announce Type: new Abstract: As Large Language Models (LLMs) advance toward embodied AI agents operating in physical environments, a fundamental question emerges: can models trained
arXiv:2610.08815v1 Announce Type: new Abstract: Agentic AI-enabled automation cannot be safely deployed in high-stakes environments on probabilistic reasoning alone. A recurring risk is epistemic drif
arXiv:2610.02765v1 Announce Type: cross Abstract: Monitoring planned driving trajectories requires accurately estimating the collision likelihood with actors whose motion is itself impacted by the ego
arXiv:2609.38479v1 Announce Type: cross Abstract: This paper presents an explainable approach to pedestrian routing, in which perceived safety is estimated from street-level imagery through an explici
Brandon Pho / Mission Local: A profile of Mariano-Florentino Cuéllar, Anthropic's first global affairs chief, who had helped shape California's first AI safety law in 2025 — In a hot and dry border to
Maximilian Schreiner / The Decoder: OpenAI prices GPT-6.1 Sol at 2/1M input and 10/1M output tokens, the same as GPT-6 Sol and Claude Sonnet 5.5, and says it performs well on safety tests — Key Points
arXiv:2609.31110v1 Announce Type: new Abstract: Large language models are increasingly used as high-level planners for mobile robots, robot manipulators, and autonomous vehicles. Recent studies show t
arXiv:2609.31524v1 Announce Type: new Abstract: Surgical scene understanding is critical for computer-assisted intervention, yet laparoscopic cholecystectomy remains challenged by the complex anatomy
arXiv:2609.29266v1 Announce Type: new Abstract: We develop a multi-label Polish content-safety classifier by fine-tuning allegro/herbert-base-cased (124M) across five categories (hate, vulgarity, sexu
arXiv:2609.26233v1 Announce Type: new Abstract: Text-to-video (T2V) services inherit their safety stack from image generation, pairing a keyword prompt filter with a per-frame checker that blocks a cl
arXiv:2609.20779v1 Announce Type: cross Abstract: Safety evaluations for large language models rely on surface-form classifiers that report declining harm scores across model generations. We provide e
arXiv:2609.19472v1 Announce Type: new Abstract: Autonomous systems increasingly rely on Large Language Models (LLMs) yet the safety infrastructure surrounding these models introduces latency and compu
arXiv:2609.04281v1 Announce Type: cross Abstract: Vision-language models (VLMs) are increasingly deployed in high-stakes settings, where a response that is reasonable in general may still be unsafe fo
arXiv:2609.00015v1 Announce Type: new Abstract: AI agents powered by large language models are evolving from isolated assistants into heterogeneous systems in which multiple agents, planners, controll
arXiv:2602.06268v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) systems are increasingly integrated into clinical workflows. However, pr
arXiv:2608.24777v1 Announce Type: new Abstract: LLM-based agents can interact with external environments through tool invocation, but this capability also introduces security risks such as file modifi
arXiv:2608.20345v1 Announce Type: cross Abstract: Conversational AI systems have become informal mental health support resources for Generation Alpha (Gen Alpha, born 2010-2024), with 13.1% of U.S. ad
the crux of why AI safety in systems that are built around LLMs is never going to work: they just can’t be trusted to follow instructions. Claude Code w/ Fable 5 has strict instructions in my project
arXiv:2608.09653v1 Announce Type: cross Abstract: Recent advances in learning-based control have enabled impressive achievements in solving complex control problems in various domains. However, since
arXiv:2501.15790v2 Announce Type: replace Abstract: Synthetic minority oversampling is typically designed and evaluated against a predictive objective, generating samples that improve downstream class
arXiv:2608.06477v1 Announce Type: cross Abstract: Computer-use agents (CUAs) face a growing threat from indirect prompt injection, where adversarial instructions are planted in the environment such as
arXiv:2608.01388v1 Announce Type: cross Abstract: Runtime safety monitors based on Linear Temporal Logic (LTL) and finite automata (FSA) are increasingly deployed to intercept unsafe tool-call sequenc
arXiv:2607.28623v1 Announce Type: new Abstract: We present PAC-MAN, a perception-aware CBF-RL framework that couples control-barrier safety with deployment-realistic onboard sensing for whole-body hum
arXiv:2607.27849v1 Announce Type: cross Abstract: An open-weight LLM can write composition setpoints every five minutes. What a plant still needs is a hard check: named constraints, logged margins, an
Google announced Gemini Robotics 2.0 on July 30 2026, launching a family of three models that enhance robot dexterity, safety, and whole‑body intelligence for humanoid machinery. The publicly released
arXiv:2607.22692v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used for emotional support despite lacking mechanisms to safely govern evolving mental health risk. Existi
Another big reason to use combination of frontier models. Chain-of-thought monitoring is treated as a reliable safety layer for agents. This DeepMind-affiliated study shows the layer can be argued out
arXiv:2607.07184v1 Announce Type: cross Abstract: Pre-deployment safety evaluations aim to inform the downstream risks of releasing a new AI model. Yet most evaluations provide limited evidence about
arXiv:2607.04645v1 Announce Type: cross Abstract: Safety alignment in large language models is typically evaluated against direct, imperative harmful requests. We show that this alignment is highly co
This is our first time telling the story of how we first built and launched Claude Code, starting with its origins in Anthropic safety research. So much more to do. We are 1% done. We've put together
arXiv:2606.31644v1 Announce Type: new Abstract: As large language models take on morally consequential roles in healthcare, legal, and hiring contexts, we need to examine whether their ethical behavio
arXiv:2606.28843v1 Announce Type: cross Abstract: Fine-tuning a large language model is a ubiquitous method for enhancing its capability on a specific downstream task. However, prior work has shown th
GPT‑5.6 Sol launches with our most robust safety stack yet. We strengthened real-time protections against high-risk cyber activity and repeated misuse, then spent weeks hardening the system with human
arXiv:2606.25782v1 Announce Type: new Abstract: With the widespread adoption of large language models (LLMs) in chatbots and everyday applications, companies increasingly need guardrails that are effe
arXiv:2606.18936v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly embedded in AI for Science (AI4Science) workflows, from scientific question answering and literature a
arXiv:2606.11949v1 Announce Type: new Abstract: We present an online monitoring system for distributional shift in deployed safety classifiers, using calibrated sequential statistics to detect when a
Anthropic: Anthropic says Claude Fable 5 uses conservative safety classifiers that trigger a fallback to Claude Opus 4.8 in <5% of sessions, in areas like cybersecurity — Today we're launching Claude
arXiv:2606.09749v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models have demonstrated impressive end-to-end performance across a variety of robotic manipulation tasks. However, these
arXiv:2606.05679v1 Announce Type: cross Abstract: Agents increasingly generate SQL, orchestrate pipelines, and automate data analysis on behalf of users. While recent work improves query correctness,
Safety by narrow control has shown to fail many times. Need more transparency on the absolute frontier, and openness close behind. I found another API that offers claude-oceanus-v1-p the pricing and t
arXiv:2512.23234v3 Announce Type: replace-cross Abstract: Infrared gas leak detection is important for industrial safety and environmental monitoring, but automatic detection remains challenging becau
arXiv:2606.03678v1 Announce Type: new Abstract: Generating safety-critical scenarios is essential for validating and improving autonomous driving systems, yet it inherently requires maximizing adversa
arXiv:2606.03460v1 Announce Type: new Abstract: Underground coal mining requires personnel and heavy equipment to operate within shared, confined, and poorly illuminated spaces where hazards such as e
glad to know Mythos' safety concerns have been addressed right as Anthropic also secured tens of billions in inference compute 👍 JUST IN: Anthropic announces it will roll out Claude Mythos “in the com
arXiv:2605.25534v1 Announce Type: new Abstract: Multimodal Large Language Models (MLLMs) excel at structural reasoning yet suffer from a sharp logical brittleness in structural consistency. We term th
arXiv:2602.12316v2 Announce Type: replace Abstract: Frontier AI systems are increasingly capable and deployed in high-stakes multi-agent environments. However, existing AI safety benchmarks largely ev
arXiv:2605.22456v1 Announce Type: new Abstract: Cloud-hosted LLM driver agents provide useful semantic judgments, but their inference latency exceeds stepwise vehicle-control windows. Learned world mo
arXiv:2605.19009v1 Announce Type: new Abstract: Humanoid robots are difficult to deploy safely because they have high-dimensional bodies, many collision constraints, and must operate near people and o
arXiv:2605.16458v1 Announce Type: cross Abstract: Image restoration models are increasingly applied to degraded medical scans, but in safety-sensitive settings they must improve image quality without
arXiv:2601.05653v2 Announce Type: replace Abstract: Existing traffic simulation frameworks for autonomous vehicles typically rely on imitation learning or game-theoretic approaches that solve for Nash
🎉 We published a new AI safety study: shopping agents fall for whimsical attacks and lose money. A whimsical attack is an absurd scenario a human would never try on another human. In one run, GPT-5.1
arXiv:2605.07764v1 Announce Type: new Abstract: Natural-language interfaces can make swarm robotics more accessible to non-expert operators, but they must translate ambiguous user intent into executab
Ever wished your agent could read PDFs, images, and Office documents as easily as plain text? Or combine the safety of a secure sandbox with the full power of Bash access? We built exactly that. Meet
Anthropic: Anthropic details how it improved Claude's safety training after finding agentic misalignment in older models, such as Opus 4 blackmailing engineers — Last year, we released a case study on
arXiv:2605.03226v1 Announce Type: new Abstract: Safety fine-tuning of language models typically requires a curated adversarial dataset. We take a different approach: score each candidate prompt's diff
arXiv:2605.00245v1 Announce Type: new Abstract: Large language models (LLMs) are now being explored for defense applications that require reliable and legally compliant decision support. They also hol
arXiv:2604.22134v1 Announce Type: new Abstract: Large Language Models (LLMs) have been widely explored in educational scenarios. We identify a critical vulnerability in current educational LLMs, pedag
arXiv:2604.19638v1 Announce Type: new Abstract: Multimodal Large Language Models are increasingly adopted as autonomous agents in interactive environments, yet their ability to proactively address saf
arXiv:2604.16980v1 Announce Type: new Abstract: Background: Large language models (LLMs) are increasingly proposed for diagnostic support, but few evaluations use real-world multimodal inpatient data,
This GitHub project (AG-X) implements deterministic safety checks for Ollama agents, designed to catch malformed JSON responses, detect prompt injection attempts, and handle model refusals before they