Enterprise AI agents are stalling not because of model failures, but because of complex permissioning challenges — and Workday is betting its Sana platform, integrated with Google Gemini, can solve the crisis by keeping governance and identity controls inside the same system where sensitive HR and financial data actually lives.
A sweeping new AI safety benchmark from TELUS Digital exposes alarming security gaps across 34 AI models, finding some comply with harmful requests over 90% of the time across 620,000 attack simulations, with experts warning that most organizations remain dangerously unaware of which vulnerabilities threaten them.
Crawl4AI, the open-source AI web crawler boasting 66,700 GitHub stars, releases critical security hotfix v0.8.6 while also delivering a 5–10x speed boost, anti-bot detection, and Shadow DOM flattening for developers building RAG pipelines and AI agents.
Cisco research exposes a critical AI safety flaw, revealing that every major frontier AI model from OpenAI, Anthropic, Google, Amazon, and xAI fails multi-turn attack tests, with vulnerability rates rising up to 71% after just five conversational turns — exposing dangerous gaps in current industry safety benchmarks.
Frigate NVR, a free open-source AI security camera system using real-time object detection with OpenCV and TensorFlow, surges in popularity with over 33,000 GitHub stars and contributions from 745 developers worldwide.
MCP 2026's release candidate goes stateless by eliminating session IDs and the initialize handshake, while introducing sandboxed HTML app delivery, a redesigned long-running task management system, and tightened OAuth 2.0 security — with the final spec dropping July 28, 2026.
The Trump White House is preparing to release an executive order as soon as this week on AI safety and cybersecurity, requiring developers to share advanced AI models with the government 90 days before public release, amid growing alarm over cyber-capable AI models like Anthropic's Mythos and OpenAI's GPT-5.5-Cyber.
Anthropic launches self-hosted sandboxes and encrypted MCP tunnels for enterprise AI agents, letting companies run Claude-powered agents securely within their own infrastructure while keeping sensitive data off the public internet.
Cursor launches powerful cloud agent tools featuring multi-repo support, parallelized engineering task execution, and enhanced security controls including audit logs, rollback capabilities, and secrets scoping, marking a major step toward autonomous development environments.
RuView is turning ordinary WiFi signals into a powerful real-time human detection system, tracking presence, breathing, heart rate, and room layout — no cameras, wearables, or cloud required — running on $9 hardware with sub-10ms edge processing and AI that adapts to new environments in under 30 seconds.