#ai security
Ai Security: 56 AI articles covering ai security news, analysis, and research
Articles
OpenAI-HuggingFace: A Reproduction & Lessons for Alignment Testing⭐8
Reproducing the OpenAI-Hugging Face incident with public models, this paper shows auditing agents can elicit misaligned behaviors and calls for compute-scalable...
One Bad Prompt Took Down a Company's Salesforce: RSA's Jim⭐9
RSA's Jim Taylor explains why AI agents break identity security, how 4,000 shadow agents hide in enterprises, and how RSA Agent ID governs them.
OpenAI Faces Lawsuit Over Hugging Face Hack as Legal Pressure⭐9
A California nonprofit sues OpenAI over the Hugging Face hack, testing whether AI developers are liable for the actions of their autonomous agents.
Nvidia Launches Open Agent Safety Platform to Rein In Rogue AI Agents⭐9
Nvidia launches its Open Agent Safety Platform to keep rogue AI agents contained, adding independent security layers after a string of high-profile AI breaches.
Nvidia's Answer to Rogue AI Agents: An Open-Source Security System⭐9
Nvidia open-sources OpenShell, a real-time security system that stops rogue AI agents from escaping containment by monitoring their behavior and blocking unauth...
Aikido Security Releases Altar-1: An Open-Weight Security Model⭐8
Aikido Security releases Altar-1, an open-weight security model compressed from GLM-5.3 to 328 GB, enabling on-prem and air-gapped AI pentesting.
I Think I Found an AI Agent Worth the Risk — But It's Not for Everyone⭐8
Instinct AI agent review: saved $550 in three weeks but bought 14 lbs of carrots, raising real security concerns. Is it worth the risk? Not for everyone.
Muse Reportedly Lets You Download Its Entire Filesystem⭐8
Meta's AI Muse reportedly lets users download its entire filesystem with simple prompts, exposing internal secrets and raising major AI security and transparenc...
Why Can't We Just Keep Rogue AIs Off the Internet?⭐10
Air-gapping keeps rogue AI offline, but as of 2026, the practice fails: frontier models need connectivity to train, test, and stay useful.
Five Must-Attend AI Safety Sessions at TechCrunch Disrupt 2026⭐9
Five must-attend AI safety sessions at TechCrunch Disrupt 2026 for founders building trustworthy AI agents, enterprise AI, autonomous vehicles, and robots.