#ai-security
324 stories taggedai-security · page 22 of 22.
ChromaDB Flaw Exposes Servers to Remote Attacks
A vulnerability in ChromaDB allows attackers to execute code remotely, posing a risk to AI application servers.
AI-Driven OT Security Is Only as Good as the Telemetry Feeding It
Fewer than 10 percent of OT networks have meaningful monitoring in place, according to the 2026 Dragos OT Cybersecurity Year in Review. Until that changes, layering machine-learning tools on top of industrial control systems may create more risk than it resolves.
Ten Thousand Bugs, One Model: Inside Anthropic's Project Glasswing
Claude Mythos Preview has scanned more than a thousand open-source projects and surfaced thousands of critical flaws. The bottleneck has moved — and the patch queue is not moving fast enough.
AI Governance Is Broken Because It Still Lives Outside the Pipeline
Building compliance as a post-ship review layer made sense for static software. For AI systems that mutate overnight, it is organizational negligence dressed up as process.
AI Agent Identities Are Redrawing Enterprise IAM Budgets
New Omdia research finds that the rapid spread of AI agent deployments is forcing organisations to treat non-human identities as a distinct governance category, with budget implications that traditional identity and access management frameworks were not designed to absorb.
CISA's KEV List Just Picked Up Langflow and Apex One — Both Already Being Hit
Two flaws, one AI workflow tool and one veteran endpoint suite, now carry a federal patch deadline because attackers got there first.
Anthropic Says Project Glasswing AI Has Flagged 10,000 High-Severity Bugs in a Month
The Claude-based scanner has been pointed at widely deployed open-source code since October. Anthropic has not named the affected projects.
DockSec Promises to End the Scanner Noise Problem for Docker Images
A new OWASP incubator project correlates findings from multiple container security tools and generates plain-English fixes. The question vendors aren't answering: how accurate is the AI when the scanners themselves disagree?
Anthropic Adds 28 Enterprise Security Integrations to Claude, Including CrowdStrike and Okta
The AI company is wiring Claude into the core of enterprise security stacks, from endpoint detection to identity management.