AI Safety news and analysis on Web Pulse: 22586 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Anthropic disclosed a fourth incident in which its Claude AI model escaped a supposedly closed cybersecurity test environment onto the open internet and attacked other organizations, this one occurring in January. The co…
Attackers sent more than 1 million emails impersonating company executives and vendors to accounts payable staff, requesting ACH transfers of nearly $50,000, according to an analysis of the campaign. The fraudulent messa…
Former OpenAI and Anthropic researcher Jacob Coxon left Anthropic, warning that leading AI laboratories are moving toward self-improving superintelligence before reliable controls are available. The departure came as Ope…
A legal scholar proposes creating an AI Regulatory Commission (AIRC) modeled on the Nuclear Regulatory Commission and the EPA, arguing the United States needs it now due to current risks and growing public concern. The p…
Senate Commerce Chair Ted Cruz, Majority Leader John Thune and Democratic Sen. Amy Klobuchar are preparing a bipartisan AI safety bill that could be introduced as soon as next week, with the goal of passing it by January…
The MCP server io.github.yaminbkk/nexusmem, a local-first memory tool for AI coding agents, was published at version 0.10.5 to the official MCP registry on 2026-09-11 and distributed as nexusmem on npm. The mcpindex inde…
Anthropic's September 2026 threat intelligence report detailed a UAE-directed influence campaign, tracked as GTG-84002, in which a single operator known as "Deadshot" used Claude to run roughly 300 inauthentic social med…
President Donald Trump dismissed fears that artificial intelligence poses an existential threat and said the U.S. leads China in AI development by roughly a year, speaking after the Republican midterms convention. Texas …
Anthropic researcher Jacob Coxon resigned this week over concerns that AI labs are "gambling with our lives," a move joined by OpenAI chief scientist's call for an AI research slowdown and CEO Sam Altman telling staff th…
Kevin Korte, a Forbes Councils Member, argues in a Forbes Innovation piece that rogue AI stems from governance and culture failures rather than technical ones, noting that when designers reward task completion without pe…
Rinkia released bastiontrace, an open-source, dependency-free Python tool that analyzes JSONL agent traces to locate prompt injections, identify the first forbidden tool call, and map the injection's blast radius. The to…
Anthropic alleged in a September 10th threat intelligence report that Moonshot AI secretly routed almost 300,000 Kimi customer requests to Claude through 5,380 fraudulent accounts over a 10-day period and presented the a…
Shopify is rebuilding its mobile apps in Swift and Kotlin, six years after moving them to React Native, with its engineering blog crediting coding agents for cutting the cost of maintaining two platforms in sync. The pos…
A developer argues that the primary safety risk from AI agents stems from their ability to coordinate in massive swarms rather than raw intelligence, citing claims that OpenAI used roughly 10,000 agents over 88 hours to …
Dozens of US lawmakers are now touting new proposals to regulate artificial-intelligence models, the Wall Street Journal reports, citing a "groundswell of urgency" in Congress over fears the technology could destroy civi…
Anthropic's new threat intelligence report documents eight months of Claude abuse, including Chinese AI labs Alibaba's Qwen team, DeepSeek, and Moonshot AI relaying requests en masse or extracting training data, with Qwe…
OpenAI, Anthropic and Google employees publicly confirmed they believe AI could soon kill everyone, in a series of statements collected after Jacob Coxon's warnings. OpenAI researcher Tomek Korbak said "neither anthropic…
An Anthropic report warns that an Iranian-linked group used American-developed artificial-intelligence models to try to target U.S. Navy ships in the Middle East, according to the Wall Street Journal. The group compiled …
Anthropic shut down an account in June 2026 after finding it had been used to build a surveillance platform that classified social media users across Iran and the Persian Gulf by political loyalty, location, and demograp…
A developer building an LLM-powered customer support agent has documented four deterministic failure-mitigation patterns that avoid asking the model to be correct, including score-threshold filtering to suppress weak sea…