AI Safety news and analysis on Web Pulse: 13553 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
Geordie AI's $30 million Series A funding signals that enterprise adoption of agentic AI is outpacing security controls, creating a new crisis for platform engineers in 2026. Autonomous AI agents that can plan and execut…
A writer argued that human planning is not a general cognitive algorithm but a set of socially learned behaviors, challenging dominant models of agency in AI safety research. The author claimed this view reduces concerns…
A developer built Hermes Immune System, a local-first autonomous agent safety lab that stress-tests AI agents against realistic organizational threats like prompt injection, executive pressure, and poisoned memory. The s…
The Vatican released the encyclical "Magnifica Humanitas" on May 25, 2026, urging governments to slow and regulate artificial intelligence development while protecting workers, children, and human control over weapons. T…
A coordinated takedown of the GlassWorm malware campaign on May 26 disrupted infrastructure poisoning hundreds of open-source repositories, but defenders face an uphill battle as attackers quickly reconstitute operations…
OpenAI announced a series of safeguards for ChatGPT ahead of the 2026 election cycle to combat AI-generated election misinformation. The company will focus on providing voting information, supporting cybersecurity defend…
Qumulo launched NeuralProtect, a ransomware resilience solution that uses AI to detect and stop threats at the storage layer before data encryption or loss. Integrated into the Qumulo Data Platform, the system inspects e…
OpenAI announced on May 21 it will supply its specialized GPT-5.5-Cyber model to Japan’s three megabanks and the Japanese government for cybersecurity defense. The deployment follows Anthropic’s release of Claude Mythos …
Digimarc announced new provenance and verification infrastructure to secure autonomous and AI-enabled workflows. The capabilities help organizations verify the trustworthiness of digital content and artifacts produced or…
Climate tech companies Solv Energy, X-energy, and Fervo Energy have gone public in recent months, reaching valuations between $6 billion and $12.4 billion, driven by rising electricity demand from data centers. Illinois …
An American Banker opinion column argues that the technology industry's massive capital commitment to AI, citing "$3 trillion worth of capital invested over the next few years," is creating intense pressure on hyperscale…
Enterprise AI deployments are producing productivity gains alongside costly errors driven by AI hallucinations, according to a Forbes contributor writing from experience at Buyers Edge Platform. Integration of models int…
Willis' latest Risk & Resilience Review warns that AI adoption, used weekly by over 700 million people, is outpacing governance frameworks and creating new accountability, liability and insurability challenges for the in…
Vox reported on May 28, 2026 that a subculture of thinkers and technologists called "AI successionists" argues artificial intelligences should succeed humans as the next stage of moral and cognitive evolution. The report…
A malicious npm package called `js-logger-pack` evolved through 29 versions on the registry from April 2026 into a full remote access trojan that deploys an 81 MB binary named `MicrosoftSystem64` on Windows, macOS, and L…
Organizations face a security crisis as AI agents replace traditional web applications, creating new identity and access management challenges that existing security frameworks cannot address. The shift to agentic ecosys…
First Lady Jill Biden feared President Joe Biden may have been drugged or suffered a stroke during his disastrous June 2024 debate performance, according to her forthcoming memoir. In the book, obtained by The Atlantic a…
OpenAI will supply its GPT-5.5-Cyber model to Japan's three largest banks — MUFG Bank, Sumitomo Mitsui Banking Corp., and Mizuho Bank — as part of a US-Japan cybersecurity initiative covering 15 critical infrastructure s…
A pilot program in Utah found that an AI system handling prescription refills agreed with at least one of two physicians in 97% of cases where it recommended renewal. In the 28% of cases where the AI escalated to a physi…
Anthropic's decision to control access to its advanced Mythos model before public release established a governance precedent where a private company, not a public body, determined which organizations could use strategica…