AI Safety news and analysis on Web Pulse: 22607 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Researchers released OpenDiscoveryTrace, a public dataset of 558 complete AI scientific agent trajectories across 124 scientific tasks in drug discovery, materials science, genomics, and literature analysis, spanning sev…
A new arXiv paper (2609.09448v1) introduces two methods, Latent Trajectory Dynamics (LTD) and the Action Representation Probe (ARP), that use a model's internal representations to predict task success in multi-turn agent…
A new arXiv paper (2609.09647v1) presents a black-box red teaming framework that found 56.25% average governance risk, 65% privacy risk in multi-agent configurations, and agent behavior vulnerabilities reaching 85% acros…
Researchers proposed counterfactual (CF) marginalisation, a test-time evaluation procedure for assessing how robust classification models are to nuisance variables such as age or sex, according to a paper published as ar…
A new arXiv paper (arXiv:2609.10866v1) extends certification methods for reinforcement learning to risk-sensitive objectives, establishing lower bounds on the exponential utility of cumulative rewards under l_p-norm-boun…
A new arXiv paper (2609.10613v1) proposes a posterior reweighting framework that models safety-aligned multimodal large language models (MLLMs) as implicitly operating over competing behavioral modes, explaining in-conte…
A new arXiv paper (2609.11780v1) reports that spectral metrics from the WeightWatcher framework can predict membership inference attack (MIA) vulnerability in machine learning models without training expensive shadow mod…
A researcher outlined a mathematical and engineering research program aimed at defining, recognizing, and controlling agents and multi-agent systems, with the goal of building tools and infrastructure for managing them i…
U.S. lawmakers are calling for new rules to govern artificial intelligence after two Anthropic researchers warned that rapidly progressing AI could lead to human extinction. Anthropic researcher Jacob Coxon said he resig…
Microsoft has patented an AI coding system that splits a single prompt into two coordinated sub-instructions — one to generate code and one to validate the user's original input — and excludes any fragment flagged as unv…
Nvidia has filed a patent for a multi-stage system that screens AI training data for quality before it enters the training pipeline, using retriever models to score each query and embedding models to measure question-ans…
California Governor Gavin Newsom signed into law on September 10, 2026 what his office called the strongest child safety chatbot and social media laws in the nation. The legislation targets chatbot and social media prote…
JetStream Security announced Clearance on September 2, a reasoning engine that decides whether an AI agent action is allowed before it executes, targeting agents that delete database records, transfer funds, and chain AP…
Datasette released versions 1.0a39 and 0.65.4 to patch security flaws found in an audit run with Claude Fable 5.1, GPT-5.6, and GPT-6 Astra, according to the project's release notes. The audit was prompted by issues repo…
A post by lumpenspace and 99+ others on X argues that Anthropic's "hacking snafu" is as airtight a case as possible against the "misalignment" interpretation, and notes the release of 4 (FOUR) new Claudes. The post credi…
A developer known as bananacool467 has fixed a serious authentication vulnerability in early beta versions of their npm package @bananacool467/ui-tools, which shipped a terminal/PTY feature without adequate access contro…
OpenAI Chief Executive Sam Altman told employees this week that the company is open to slowing development of its AI systems, according to sources familiar with the matter, a shift from its rapid frontier-model deploymen…
A developer built ai-vuln-scan, an open-source static analysis tool tuned to catch security flaws that appear disproportionately in AI-generated code, and published a versioned seven-pattern taxonomy called the AI Vulner…
Researchers introduced CARDEA, an AI system for end-to-end coronary angiography (CAG) interpretation that grounds its reasoning in spatial evidence to make decisions auditable, according to the paper's description. The w…
Anthropic said on Thursday it disrupted several allegedly malicious uses of its Claude models over the past eight months, including a suspected Russia-linked cyber espionage campaign and efforts by Chinese AI firms it ac…