Tracebit

Most ISO 27001 audits come down to two questions: does the control exist, and is it actually working? Here's a detailed, bounded mapping of Tracebit's deception-based detection to the 2022 Annex A control set - 12 controls with direct evidence, plus suppo

At Black Hat, OpenAI named honey tokens and deception as a key defensive investment against offensive AI agents. We map the publicly disclosed TTPs from the OpenAI / Hugging Face incident against Tracebit deception, stage by stage.

AI agents can now run complex cloud attacks on their own, escalating from a single low-privilege key to full admin within minutes. We sat down to walk through Tracebit's latest research on stopping them - watch the full recording on YouTube.

I recently joined the AI Security Podcast with Ashish Rajan and Caleb Sima to discuss what deception means in a world of AI agents. Listen now.

We're shipping our AI-attacker research as product. Context Bomb Canaries for AWS stop an a malicious agent mid-attack by tripping its own guardrails, and the new AI Posture view lets you tell your sanctioned coding agents apart from an attacker's automat

We planted context bombs, short strings that trip an AI model's own safety guardrails, inside canary secrets in a live AWS environment. Across 5 frontier models and 152 runs, they cut successful attack paths from 91% to 15% - stopping attackers outright a

Tracebit introduces

Episode 5 features Ollie Whitehouse, CTO of the UK's National Cyber Security Centre, on 23 years of running deception and building NCSC's national-scale cyber deception programme.

Today we're extending detection of an intrusion to the earliest point: the perimeter, the moment stolen access is first used. Introducing Perimeter Sensors - believable canary services on your own domain, bound to canary credentials seeded across your org

AI agents can compromise a cloud environment in minutes. We sent ten frontier AI models to attack a live AWS cyber range - across 951 runs, AI reached admin in an average of 14 minutes, but canaries warned the defender first in 95.9% of compromising runs.