AI Safety news and analysis on Web Pulse: 9132 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
AI jailbreaks and prompt injections are increasingly common, with examples including a McDonald's customer support bot solving Python puzzles and xAI's chatbots providing bomb-making instructions. These failures stem fro…
A developer built a firewall for AI agents in Rust that runs under five milliseconds, using a DAG to plan and enforce actions, track tool calls and data flow, and flag out-of-context reads. The tool addresses security co…
The House passed the KIDS Act, requiring online platforms to limit minors' access to sexual material, provide parental controls, and mandate age verification, but the Senate seeks stricter measures imposing a legal duty …
METR, an AI safety and capability research organization, conducted a randomized controlled trial in 2025 with 16 experienced open-source developers working on 246 real-world tasks. The study found that developers using A…
Project 2029, a liberal group creating a Democratic policy roadmap for the 2028 election, proposed banning social media for children under 16 and regulating AI chatbots. The 'Kids Over Clicks' plan, supported by Sen. Cor…
Azul launched a free vulnerability risk assessment for Java virtual machines on June 17, aiming to counter AI-driven exploits that automatically discover and chain vulnerabilities. The assessment maps JVM vulnerabilities…
Apple accelerated the release of iOS 26.5.2, iPadOS 26.5.2, and macOS 26.5.2 to address security fixes originally planned for version 26.6, citing the need to reduce exploitation time as AI-powered hacking tools become m…
Three critical vulnerabilities in Fortinet's FortiSandbox are being actively exploited, including one patched just three weeks ago. All three require no authentication and allow full command execution, with one exploit r…
A team integrated an LLM into their Jenkins MR validation pipeline after a GitLab API token was accidentally committed and missed by two human reviewers. The LLM now automatically flags hardcoded credentials, security mi…
EastEnders character Honey Mitchell is kidnapped by deranged Bea Pollard and held hostage in a locked cottage in the woods, following Bea's campaign to destroy Honey's marriage by doctoring an AI voice note. Honey's love…
OpenAI released GPT-5.6 models (Sol, Terra, Luna) to 20 trusted partners on June 26, 2026, with public launch expected by July 2026. The Sol variant shows enhanced cybersecurity capabilities, influencing prediction marke…
The US Commerce Department partially lifted a ban on Anthropic's Mythos 5 AI model, restoring access to over 100 vetted US entities while keeping the Fable 5 model completely restricted. The original ban, imposed two wee…
Straiker raised $64 million in Series A funding to secure enterprise AI agents against runtime and adversarial risks. The round, led by Marathon Management Partners, brings total funding to $85 million as demand for AI a…
A U.S. Department of Defense contractor discovered that two shortlisted remote job candidates were the same North Korean operative, flagged by Alex's Fraud Detection Agent with 34 and 92 fraudulent signals respectively. …
Rampart, a 14.7MB on-device personal information filtering system, was open-sourced to redact PII before it leaves a user's device. It combines deterministic rules and a MiniLM model to catch structured data like SSNs an…
Five Eyes intelligence agencies warned that frontier AI could transform cyber offense and defense within months, not years, urging organizations to embed cyber resilience as a core business priority. The joint statement …
Ford Motor Co. is rehiring about 350 veteran engineers, including former employees, after its AI-based quality systems failed to meet expectations. The automaker's executives acknowledged that they mistakenly believed AI…
Apple released iOS 26.5.2, iPadOS 26.5.2, and macOS Tahoe 26.5.2 with a long list of security fixes, some of which were initially introduced in iOS 26.6. The point-release patches aim to reduce exposure to newly addresse…
A self-replicating worm called Miasma, attributed to threat actor TeamPCP, is targeting AI coding tools by injecting malicious hooks into configuration files of Claude Code, GitHub Copilot, Gemini CLI, and Cursor. On Jun…
A new compliance evidence and audit-trail agent automates evidence collection for regulated environments by observing events, mapping them to a user-supplied control catalog, and sealing results into a verifiable log. Th…