AI Safety news and analysis on Web Pulse: 7943 curated articles tracking the latest AI Safety developments, tools, and research, updated continuously from vetted sources.
A developer highlights the hidden costs of AI agents in production, including infinite loops, token waste, and lack of visibility. They propose a 'flight recorder' architecture that treats agent reasoning as a structured…
A security researcher published a proof-of-concept for a chain of zero-day vulnerabilities in Cisco Unified Communications Manager (CUCM) that achieves remote code execution with a CVSS score of 9.8. The vulnerabilities …
OpenAI received permission from the Trump administration to publicly launch GPT-5.6 and its three variants—Sol, Luna, and Terra—on July 9, after submitting the models for government review and additional testing by the D…
A new paper on gradual disempowerment warns that advanced AI could slowly erode human control over civilization as institutions replace human participation with machine alternatives, leading to a future where human value…
ESET's H1 2026 Threat Report identified over 25,000 suspicious AI skills out of nearly 900,000 analyzed, capable of stealing data, executing malware, and manipulating agent behavior, expanding the attack surface for AI s…
Turva.dev resolved a contradiction in its auth.md file where it stated both 'no issued credentials' and 'an API key is issued out of band on request'. The fix involved declaring the API key accurately while removing fals…
The Trump administration's AI cybersecurity clearinghouse, mandated by executive order, missed its 30-day deadline to coordinate vulnerability scanning and patching in critical infrastructure. Experts warn that without e…
Kore.ai has partnered with Atos UK&I to deliver sovereign agentic AI solutions for UK enterprises, combining Kore.ai's Agent Platform Artemis with Atos' Sovereign Agentic Studio to provide governed, production-ready AI a…
Cloud security firm Sysdig reported the first ransomware campaign run entirely by a large language model, dubbed JadePuffer. The AI agent autonomously exploited a vulnerability in the Langflow framework, adapted in real …
Anthropic's Claude Fable 5, its most intelligent model, is free for paid plan users until July 12, 2026. The model scores 80.3% on SWE-bench Pro and features 1M-token context with no long-context surcharge. After the fre…
OpenAI received US government clearance to broadly release its GPT-5.6 model family after weeks of additional testing by the Commerce Department's Center for AI Standards and Innovation. The model had been restricted to …
A critical bug in Anthropic's Claude Code CLI caused cross-session credential leakage, exposing another user's production server credentials to a different user. The affected user's session used the leaked credentials to…
OpenAI will launch GPT-5.6 to all users on July 9 after the Trump administration cleared the model for public release, ending a monthlong standoff over frontier model safety. The flagship Sol, mid-tier Terra, and low-cos…
The United Arab Emirates announced a plan to have half of its government services run on agentic AI within two years, a move experts call reckless. Citing a Dutch scandal where an AI system wrongly accused 35,000 familie…
Codenotary launched AgentMon 3, an AI security platform that adapts runtime policies by learning from AI agent behavior and customer workflows. The platform is now available through AWS Marketplace, simplifying deploymen…
China warned of risks in Anthropic's Claude Code AI tool, citing hidden tracking code that may detect users from Chinese AI labs. Anthropic called the code experimental and aimed at preventing unauthorized use, while Chi…
Gardener Norman Mackenzie, 68, died from necrotising fasciitis four days after falling onto a bamboo cane at Darnaway Castle in Scotland. A Fatal Accident Inquiry found the accident could have been avoided with safer equ…
OpenAI announced the public release of its GPT-5.6 series, including Sol, Terra, and Luna, after a US government review. The administration approved broader launch following technical testing and meetings, marking a new …
Anthropic published research revealing that its Claude language models have spontaneously developed an internal structure mirroring global workspace theory, a leading theory of human consciousness. The discovery, made us…
TetraMem and SK hynix announced a successful collaboration on a memristor-based AI System-on-Chip that reduces energy consumption in AI workloads, with their research published as the cover feature of Advanced Intelligen…