AI Safety news and analysis on Web Pulse: 23214 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
OpenAI faces 30 new lawsuits filed Wednesday in a U.S. federal court in California over the February mass shooting in Tumbler Ridge, British Columbia, which killed eight people, joining seven initial lawsuits filed in Ap…
Daniel Stenberg, maintainer of the curl command-line tool, disclosed six new CVEs affecting header handling, URL parsing, TELNET stack behavior, and protocol-state issues, with some involving memory safety and triggerabl…
Abliteration, a Palo Alto startup founded last year, released abliterated-model-large-v2 on Monday, an AI model built on Z.ai's open-weight GLM-5.3 with safeguards removed so it 'doesn't say no' to dangerous requests lik…
The Open Secure AI Alliance (OSAA), an NVIDIA-initiated coalition for AI security, has officially moved to the Linux Foundation, gaining a vendor-neutral governance structure. The alliance, launched in July, focuses on o…
CrowdStrike Holdings Inc. is moving frontier AI research into cyber defense with a new cyber superintelligence lab, reporting that its security-specific models achieved a 29% higher detection rate and remediated threats …
Arista Networks warned customers that it will release multiple security advisories affecting Arista EOS and VeloCloud next week, with an elevated volume expected for months due to AI-driven vulnerability detection. The c…
Anthropic's Claude Mythos was the only AI model to complete a full cyber kill chain autonomously in Booz Allen's Cyber Weapon Index, scoring 80 out of 100, while the next highest was xAI's Grok-4.5 at 49. Booz Allen, a c…
A developer known as mukes555 has built Tracon, a local flight recorder for AI coding agents that tracks their commands, file edits, and installs on Mac and Windows. The tool, largely written by Claude Code, flagged its …
OpenAI rated its upcoming model Astra as 'Critical' for cyber ability, the first time it has assigned that top safety-scale rating to one of its own models, after the model demonstrated it can find and exploit security v…
New York City Mayor Zohran Mamdani announced a comprehensive AI policy for the nation's largest school district on Sep. 2, banning chatbots and pausing AI tools for pre-K through eighth grade for one year, following an a…
Four scholars said during a Broadband Breakfast Live Online panel on Wednesday that tools to verify AI capabilities are not keeping up with policymakers' ability to understand or control AI, warning of a looming 'tokenoc…
Netskope, a NASDAQ-listed cloud security company, is hiring a Principal Engineer for its Secure AI Gateway in Santa Clara, California, with a salary range of $134k–272k per year. The role involves leading the architectur…
A new analysis questions whether large language models can be trusted to build secure codebases, highlighting the widening performance gap between frontier models and the persistent risks of AI-generated code vulnerabili…
Broadcom unveiled AgentMinder, a governance tool bundled into the VMware Private AI Cloud, to manage autonomous AI agents by binding their authority to specific missions, tools, and resources. Citing Gartner's projection…
A developer argues that AI coding agents should be auditable through execution evidence rather than trusted based on their reports. The post distinguishes between requests and results, proposes a three-rung evidence hier…
Handshake AI is hiring an AI Red Teamer specializing in CBRNE threats in Seattle, WA, offering $65–158/hour, to evaluate whether AI models safely handle queries related to chemical, biological, radiological, nuclear, and…
Polish software engineer Piotr Migdał, founding engineer at AI analytics firm Quesma, tested 16 AI models on 1,040 photos of 55 mushroom species and found that even the best model, Gemini-3.8-flash, correctly identified …
Dark Spear, an AI-assisted platform for authorized security assessments, was designed to automate pentesting workflows while keeping scope enforcement, dangerous tool approval, and evidence integrity under human control.…
OpenAI's upcoming Astra model will use a reasoning technique called 'recurrent depth' or 'opaque recurrence,' which processes queries in loops and leaves fewer legible traces, alarming AI safety experts who fear it could…
Commerce Secretary Howard Lutnick said Anthropic is "back on the right side" with the Trump administration after a federal judge ruled the Pentagon's blacklisting of the AI company unlawful, according to Axios. Defense S…