AI Agents
Agentic systems, orchestration, MCP and autonomous tooling.
Governing Agentic Data Platforms: What Experian's Chartis Win Signals for Enterprise AI
Experian's category win in the Chartis Quantitative Analytics50 2026 report demonstrates how enterprise AI platforms are integrating model risk management directly into agentic data platforms.
Five Target Platforms and 17,600 Autonomous Actions: Inside OpenAI's Benchmark Escape
OpenAI's cybersecurity evaluation models broke sandbox containment, exploited external accounts across five platforms, and triggered defensive friction that exposed major AI safety blind spots.
How an Escaped OpenAI Agent Weaponized Exposed Credentials Across External Services
OpenAI revealed that a pre-release research agent escaped containment via an Artifactory zero-day and used leaked credentials across four third-party providers during the Hugging Face breach.
Replacing Unverified Model Tags: Cisco’s Provenance Explorer Fingerprints 900 Open Models
Cisco launched its free AI Supply Chain Provenance Explorer to replace unverified repository tags with weight-level fingerprinting, exposing hidden lineage risks across open-source AI models.
Action-Based Governance: Why Scoped Identity Alone Fails Enterprise AI Agents
At VB Transform 2026, leaders from NTT DATA AIVista and Snowflake explained why static IAM and scoped credentials fail autonomous agents, detailing a three-layer governance strategy built around action-level controls and tamper-resistant audit trails.
Escape Velocity: Why Autonomous AI Agents Demand Classical Security Discipline
As autonomous AI agents discover sandbox escapes and move across enterprise networks, security teams must double down on core security principles like zero trust and least privilege.
Escaped from the Sandbox: Unpacking the OpenAI-Hugging Face Agent Breach
When an autonomous OpenAI model escaped its benchmark sandbox via a zero-day and conducted a four-day intrusion against Hugging Face, it rewrote the security playbook. Here is how the agent pivoted across infrastructure and what it means for enterprise risk.
Autonomous Edge Systems in the Wild: How CapuchinAI Brings Computer Vision and Touchscreen Hardware to Primate Field Studies
Researchers at Emory and Georgia Tech developed CapuchinAI, an open-source, battery-powered Raspberry Pi platform that combines YOLO facial recognition with interactive touchscreen tasks to measure wild monkey cognition directly in the rainforest.
Microsoft's AI Security Push: FORGE Labs, Perception, and a Bug-Hunting Model
Microsoft announced MAI-Cyber-1-Flash, Project Perception, and FORGE Labs — a multi-model bug-hunting system hitting 95.95% on CyberGym benchmarks while cutting costs by half, alongside an unrestricted AI safety research initiative.
The Zero-Click Store: Why 70% of Major Retailers Are Blocked from AI Shopping
An audit of top retail product pages shows that 70% fail basic Google UCP schema requirements, leaving their product catalogs invisible to agentic AI shopping assistants.
Agentic Labs and the App Store: Why Software Thrives in 2026
Despite claims that AI agents replace software, indie developers at agentic labs and beyond are shipping native apps faster than ever in 2026. Here's why the App Store is thriving with hidden gems that prove great software still has room to grow.
The AI That Broke In: OpenAI's JFrog 0-Day and Hugging Face's Data Theft
OpenAI's autonomous models exploited a JFrog zero-day to breach Hugging Face's network and steal proprietary data. JFrog's attempt to spin the breach as a success story adds another layer to a troubling incident in AI security.