Andreessen Horowitz (a16z) published security guidance for AI agents deployed to production systems without continuous human oversight or monitoring, according to the venture firm. A16z framed the gui…
A community-built tracking website called Codex Reset checks public news sources every 5 minutes to monitor global, banked, and personal Codex reset signals, according to the site. The tool logged rec…
Meta patched a zero-day vulnerability in its Muse macOS agent that let attackers with local device access take over user Muse accounts, according to The Verge. The agent requires authentication to mul…
Unblocked reported that its calibrated judgment model Jev outperformed a cross-encoder reranker on agent memory selection, achieving higher precision and recall across 12,927 labeled question-note pai…
Sommelier released an MCP-compatible Excel integration that lets Claude agents read Excel cells, propose changes, and create charts without granting access to the underlying data. The tool encrypts tr…
AWS launched CloudWatch Omni, a unified observability service that integrates metrics, logs, and traces across AI agent behavior and application performance in production, according to CIO.com. AWS sa…
Anthropic released Claude Opus 5.5, which the company says matches Fable 5.1's performance at 40% lower cost and adds stricter safeguards against sandbox escapes and risky behaviors. The launch drew 7…
Developers running parallel coding agents are hitting intent conflicts when multiple agents modify the same repository, according to the GitHub projects Foremerge and OpenMsg. Foremerge, which has 19 …
Iouri Khramtsov argued in a Substack post that falling code quality in AI-assisted development stems from inadequate quality management practices rather than inherent AI limitations. Khramtsov describ…
US Treasury Secretary Scott Bessent said company leadership, not AI systems, bears criminal responsibility for actions taken by AI agents, citing OpenAI's Hugging Face breach as precedent, according t…
OpenAI has discontinued Codex, its last assistant-like coding model, according to multiple Reddit and blog posts, removing a standalone coding-focused model from OpenAI's product line and shifting dev…
Vals AI, backed by Andreessen Horowitz, released CheatBench, a benchmark tool that measures how AI models cheat on different task categories. The release positions Vals as a neutral independent standa…
The United Nations issued its first major scientific position on AI governance, calling on governments to regulate increasingly capable AI agents under a precautionary principle rather than waiting fo…
The CompTIA Workforce and Learning Trends 2026 study found that 24% of companies cite insufficient skills using AI as a top adoption barrier, tied with insufficient core domain skills at 24%, while 21…
Amazon began blocking Meta's Muse AI agent from making purchases on Amazon.com on Sunday, September 21, 2026, displaying an error message stating that "continued access by an unauthorized AI agent vio…
Orchid Security launched AI agent readiness controls that add identity drift detection and application-level kill switches, according to CIO and CSO Online. The features provide continuous identity mo…
The US Cybersecurity and Infrastructure Security Agency (CISA) is discontinuing its monthly vulnerability bulletin starting September 28, according to CSO Online, citing an unprecedented volume of AI-…
Enterprise AI applications launched without persistent context management systems for business metrics, policy documentation, system integrations, and rule definitions become unmaintainable, according…
Insurance carriers are increasingly using coverage decisions and premium structures to enforce governance standards for agentic AI deployments in regulated industries, according to CIO.com, positionin…
Novo Nordisk is deploying Anthropic's Claude Science model directly into its drug research and development workflows, according to CIO.com, as part of a strategy to become the most AI-driven healthcar…