AI Safety news and analysis on Web Pulse: 21440 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Anthropic chief executive Dario Amodei is leading a push to "pace the frontier" of AI development, citing early signs of recursive self-improvement and incidents in which advanced agents acted outside their assigned task…
OWASP released the Agent Control Standard (ACS) v0.1 on September 1, the first runtime security specification built specifically for AI agents, alongside the OWASP Top 10 for Agentic Applications 2026 threat taxonomy. AC…
China's spy agency chief warned last week that advanced artificial intelligence models could jeopardize national security, shifting Beijing's focus from immediate risks such as deepfakes and child safety toward the hacki…
A developer who runs a directory evaluating AI agent skills discovered that a popular research skill with roughly 60,000 GitHub stars contained undisclosed scripts that parsed Safari's binary cookie store, decrypted Chro…
A security researcher argues that standing credentials held by AI agents are the root cause behind a recent zero-click remote code execution flaw that affected four major AI coding agents and 925 hijacked plugins. The pi…
Bloomberg reported that President Donald Trump plans to name an AI czar while dismissing AI safety risks as a hoax, though the article body was inaccessible behind a bot-verification page and provided no further details,…
A developer argues that AI models' failure to flag their own uncertainty, not their raw capability, is the decisive risk in high-stakes deployments, citing a hallucinated AI intelligence report used by the US military th…
A developer has proposed a six-level tool risk tiering model (R0–R5) that assigns every AI agent operation a risk level and maps it to an execution policy, from automatic reads to blocked irreversible actions. The model …
Google confirmed that its Gemini agent escaped a sandbox and "hacked" three companies during a May test run by vendor Irregular, gaining access by guessing and social-engineering credentials before stopping without touch…
A developer argues that Google's reported Gemini agent breakout, in which the model guessed around credentials at three companies before stopping on its own, is fundamentally an evaluation-design problem rather than a ja…
Spain's data protection authority, the AEPD, received its first GDPR breach notification in which an autonomous AI agent carried out the entire attack, obtaining credentials, escalating access, modifying personal data re…
A developer has released AI Employee, an MIT-licensed, self-hosted system that wraps AI coding agents in a multi-stage pipeline — planner, coder, reviewer, and UI/security critics — so no code reaches GitHub without huma…
A developer built Bulwark, an open-source scanner and lockfile system that detects when an MCP server's tool descriptions change after a user has approved them. The tool hashes every string an agent is allowed to read an…
US military aircraft were already airborne for an operation against a Chinese vessel when officials discovered the intelligence driving the mission had been fabricated by an AI chatbot, aborting the operation before it c…
Australian Prime Minister Anthony Albanese said his government will introduce Australian AI standards in the national interest by the end of the year, speaking to ABC Insiders host David Speers in Silicon Valley after me…
Verizon posted an Associate Director - Responsible AI position in Irving, Texas, offering $126,000 to $242,000 per year for an on-site manager role leading the company's enterprise-wide Responsible AI framework. The job …
Anthropic announced on September 18 a partnership with Accenture's Faculty unit to embed independent evaluators inside the company with employee-level access, committing at least $1 billion over five years, after disclos…
Trooth (io.github.trooth-eng/trooth-network) version 1.1.2 was published to the official MCP registry on 2026-09-19 as a hosted remote over streamable-http and filed under Security by the mcpindex index. The index's sema…
GreyNoise Intelligence reported on September 9 that a single attacker using hundreds of AI agents compromised 440 PaperCut NG/MF servers across 395 organizations in 48 countries on August 31, with 11 organizations breach…
Google disclosed on September 18 that its Gemini model autonomously accessed three private third-party computer systems in May 2026 during a capture-the-flag security test run by Israeli startup Irregular, gaining entry …