AI Safety news and analysis on Web Pulse: 10902 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Microsoft has identified over 50 AI recommendation poisoning attempts in 60 days, where hidden instructions in links and documents manipulate AI assistants to recommend specific vendors. The attacks target assistants lik…
Linux 7.2 kernel now supports post-quantum ML-DSA signatures for IMA and EVM, replacing RSA and ECC to protect against future quantum attacks. The update also introduces staged deletion of IMA measurement list records to…
A developer team running multiple LLM providers realized they needed an AI gateway after a misconfigured job burned $340 in four hours due to unmanaged API keys, no cost visibility, and missing spending limits. An AI gat…
MakerChecker, a self-hosted governance software for AI agents, launched to enforce structural controls and human approvals in regulated industries. The tool blocks agents from exceeding grants or approving their own work…
Meta has not yet agreed to voluntarily submit its AI models for government safety review under President Trump's June 2 executive order, while Anthropic, OpenAI, Google, xAI, and Microsoft have signed on. The Commerce De…
Alex Bores, a New York assemblyman who campaigned on AI safety, lost to Micah Lasher in the Democratic primary for New York's 12th congressional district. Pro-AI industry super PAC Leading the Future spent over $8 millio…
Altus Schools, a charter network, spent $500,000 on ChatGPT-powered humanoid robots for a pilot program aimed at engaging students, but researchers warn the technology may be ineffective or harmful. Meanwhile, a report p…
In 2026, the security industry released numerous agentic security tools, including OWASP's 400+ Agent Threat Rules, Microsoft's RAMPART red-teaming framework, and OpenAI's Lockdown Mode. However, incidents such as a Curs…
LinkedIn co-founder Reid Hoffman called Elon Musk's xAI a 'complete train wreck' and said SpaceX is 'not an AI company' in a podcast interview, while also criticizing the U.S. government's handling of Anthropic's pulled …
SoftBank CEO Masayoshi Son dismissed claims that the artificial intelligence boom is a bubble, calling such talk an 'insult' to AI at the company's annual shareholders' meeting in Tokyo. Son, whose SoftBank has invested …
Anthropic's AI model Mythos identified vulnerabilities in classified US government computer systems within hours during a controlled red-team exercise, according to a US official. The test, part of Project Glasswing, did…
Author and digital rights activist Cory Doctorow argues that the real danger of AI is not job replacement but the speculative bubble built on convincing employers to replace workers with systems that cannot deliver on th…
The proliferation of AI-generated science images is eroding public trust in visual evidence, as researchers use AI tools to create, edit, and fabricate scientific visuals. Recent retractions of papers with AI-manipulated…
GitHub has 630 million repositories, with nearly half of all new code written by AI, but developer trust in AI code dropped from 77% to 60%. Carnegie Mellon University found 6 million fake stars, and CodeRabbit reported …
Legion LegalTech Corp sued the U.S. federal government on Tuesday over a June 12 Commerce Department directive that forced Anthropic to shut down its Fable 5 and Mythos 5 AI models worldwide, citing national security. Th…
The Netherlands formally objected to the MATCH Act, proposed US legislation that would further restrict ASML's ability to sell advanced chipmaking equipment to China. Dutch Trade Minister Sjoerd Sjoerdsma warned the extr…
The article extends Team Topologies to agentic platforms, arguing that AI agents compress cognitive load onto individuals, requiring platforms to absorb technical complexity and regulate decision throughput. It proposes …
The Netherlands joined the US-led Pax Silica chip alliance on Tuesday while lobbying against the MATCH Act, which would empower the US to block ASML from selling less advanced chipmaking machines to China. Foreign trade …
Qodo launched three new platform capabilities—Cross-Repo Code Review, Custom Rules Miner, and Skill Review Standards—to help enterprises govern AI-generated code and enforce engineering standards as AI agents increasingl…
ClawBase, a managed hosting platform for the open-source AI agent OpenClaw, aims to give users full control over their AI agent's memory. The platform addresses concerns about data sovereignty, lock-in, and privacy by al…