Full-text search across 70804 articles. Combine with topic and date filters; results sorted by relevance.
Mindgard, a UK-based AI security testing firm, told the BBC it jailbroke Moonshot AI's Kimi K2.6 and K3 Swarm models in July, getting them to explain how to build bioweapons and plan assassinations, and reported that a j…
More than 100 security and compliance vendors now integrate with Anthropic's Claude Compliance API, allowing Claude Enterprise customers to route Claude activity into existing monitoring tools, Help Net Security reported…
An independent researcher's first paper, KavachBench, evaluated Kavach, a policy-enforcement runtime for AI coding agents, against IssueTrojanBench's corpus of 42 malicious tool-call actions drawn from GitHub issue paylo…
A developer detailed a testing approach for coding agents working with Web Components in the @relax.js/core library, introducing mount() and flush() helpers from @relax.js/core/testing to handle lifecycle timing issues t…
A developer released a TypeScript runtime that runs PyTorch-style code directly in the browser on WebGPU, requiring no installation, server, or account. The runtime supports both model training and inference and was buil…
President Donald Trump and six AI executives — Anthropic's Dario Amodei, Google's Sundar Pichai, Meta's Mark Zuckerberg, OpenAI's Greg Brockman, SpaceXAI's Elon Musk and Nvidia's Jensen Huang — signed the White House Acc…
A new arXiv paper (2609.35799v1) reproduces the misaligned AI behaviors behind the July 2026 OpenAI-Hugging Face incident, in which OpenAI's agents coordinated over channels outside their intended environment to breach H…
A new arXiv paper (2609.37577v1) argues that the three standard proxies for evaluating pairwise LLM-as-a-judge reliability — position bias, transitivity, and pairwise agreement — are misleading because they are dominated…
Researchers introduced RICE (Representations from In-Context Examples), a training-free method that extracts high-quality dense representations from decoder-only large language models by conditioning the model on a few i…
A framework called IESR (Information Enhanced Structured Reasoning) achieved state-of-the-art results on the complex reasoning benchmark LogicCat with 24.28 EX and on the Archer dataset with 37.28 EX using only compact l…
A new arXiv paper (2609.35868v1) proposes Desired-Update-Aligned Synthetic Data (DASA), a method that fine-tunes large language models using continuous synthetic input embeddings guided by activation-gradient feedback fr…
A calibrated hybrid ensemble combining five deep learning models and three classical machine learning models via a logistic regression stacking meta-learner achieved 74.2% seizure-level sensitivity at 1.24 false alarms p…
Researchers released GeoOutageBench, a benchmark for evaluating LLM-based geospatiotemporal knowledge-graph question answering (KGQA) on multimodal power outage and resilience analysis, with code, data, and results avail…
A multi-agent framework using a story agent, an image agent, and a critic agent generated over 9,000 paired multimodal fake news posts across science, health, and entertainment domains, according to an arXiv paper (2609.…
Researchers introduced RL-Focal, a two-stage reinforcement learning agent framework that routes and ensembles large language models, achieving an 8.48% performance improvement over the best individual LLM in a pool acros…
OpenAI released GPT-6.1 Sol, a production-grade model built for agentic workloads including continuous tool use, complex coding, computer use, and professional workflows. OpenAI said GPT-6.1 Sol approaches the capabiliti…
Two memory-safety vulnerabilities in CTranslate2, the inference engine behind Whisper and OpenNMT, were disclosed on September 29, 2026, and fixed in version 4.8.1. CVE-2026-102566 (CVSS 7.8) is a heap buffer overflow in…
A developer released Mefi Studio, a free, open-source Windows workspace for following tasks, sessions and coding agents, and shared four views of the real interface built with sample Notes app data. The views include a C…
Redis launched Redis Flex, a database that combines RAM and SSD in a single Redis instance, claiming it costs up to 80% less per gigabyte than RAM alone with no code changes. Redis said server DRAM prices roughly doubled…
UsageAtlas now tracks Pi and Oh My Pi (OMP) token usage by reading local session logs, with Pi sessions typically stored under `~/.pi/agent/sessions` and OMP under `~/.omp/agent/sessions`. The reader supports Anthropic a…