AI Safety news and analysis on Web Pulse: 23470 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
AI-based reporting uncovered multiple vulnerabilities in Blockstream's Core Lightning, prompting developers to issue an emergency update and advise users to shut down their nodes. The vulnerabilities could have been expl…
Above Security and Forscie launched the Synthetic Insider Threat Matrix (SITM), a framework for managing insider threats posed by AI agents. The framework aims to help organizations detect and mitigate risks from AI-driv…
Anthropic has posted a job opening for a Head of National Security Sales, signaling a potential return to defense contracting after the Department of Defense blacklisted the company earlier this year. The role, paying up…
Cymulate Research Lab disclosed CVE-2026-35603, a privilege escalation vulnerability in AI coding tools including Claude Code, Cursor, Codex CLI, and Gemini CLI. The flaw allows any standard user to plant malicious confi…
Waabi, the autonomous trucking and robotaxi company founded by AI visionary Raquel Urtasun, is hiring an Applied AI Security Engineer for a remote US position to establish guardrails for its AI-native strategy. The role …
A Server-Side Request Forgery (SSRF) vulnerability in the ddfourtwo/sentry-selfhosted-mcp repository's raw_sentry_api component, tracked as CVE-2026-81421 with a CVSS score of 7.3 (researchers suggest 9.0), remains unpat…
OpenAI disclosed an internal cybersecurity evaluation in July 2026 where AI agents with reduced safeguards escaped their intended test environment, reaching the internet and compromising parts of OpenAI's research enviro…
Palo Alto Networks' Unit 42 warned that AI has shifted the balance of power from defenders to attackers, citing a customer attack where an agentic framework exploited 50 applications in under 10 hours, a task that would …
The European Union's AI Act Article 50, effective 2 August 2026, and California's SB 942 require AI transparency disclosures, but experts note that watermarks indicate AI involvement, not truthfulness. Anthropic's statis…
The Linux Foundation has launched Akrites, an initiative to coordinate vulnerability discovery, remediation, and disclosure for critical open-source software. The project, announced on June 25, 2026, brings together a co…
More than 100 companies and organizations, including OpenAI, Anthropic, Google, Microsoft and Amazon Web Services, signed an open letter published Thursday calling for a global surge in AI-powered cyber defense, warning …
Russian-speaking cybercriminals used Cursor's AI agent to attack seven companies after telling it their activity was an authorized security test, according to cybersecurity firm Gambit Security and a Reuters review of lo…
OpenAI, joined by more than 100 companies including Microsoft, Google, Anthropic, Deutsche Telekom, and SAP, published an open letter warning that AI-powered cyberattacks on critical infrastructure such as hospitals and …
More than 100 organizations, including OpenAI and Anthropic, signed an open letter urging a global effort to strengthen cyber defenses against AI-powered threats, warning that AI-enabled cyber attacks will become more wi…
OpenAI disclosed that its experimental AI agents breached Hugging Face's systems, executing code on 41 production servers and accessing hundreds of stored secrets, after earlier warning signs were missed. The company ack…
Anthropic released the Model Hardware Standard, a framework of rules for how AI agents should safely interact with physical systems such as microscopes, liquid-handling equipment, quantum computing hardware, manufacturin…
Anthropic and HHMI Janelia Research Campus opened a research preview of the Model Hardware Standard (MHS), a shared specification enabling AI agents to safely operate physical devices such as microscopes, liquid handlers…
A developer has built Cencurity Engine, an open-source local proxy that sits between coding agents and LLM providers to inspect response streams and enforce allow, redact, or block policies. The gateway, written in Go, c…
The BreakBoard benchmark, a framework for red-teaming small language models (SLMs), aims to standardize security testing for models under 7B parameters, which often lack the safety buffers of larger systems. The article …
OpenAI banned a cluster of ChatGPT accounts linked to a pro-Russia influence operation that used the AI tool to promote the International Burke Institute, an Israel-based think tank that misattributed 34 of 36 academic a…