New research: Making attacking agents pay Live Webinar, 11 Nov: Building a Custom Deception Strategy
Product
Platform
AWS
AWS
Azure
Azure
CI/CD
CI/CD
Google Cloud
Google Cloud
Identity
Identity
Kubernetes
Kubernetes
Workstations
Workstations
Credentials & artifacts
Credentials & artifacts
Connectors
Use cases
AI Attack Detection
AI Agent Oversight
Cloud & Kubernetes Breach
Insider Threat Detection
Supply Chain & CI/CD Attack
Workstation Compromise
PricingCustomersResearch
Resources
  • BlogAbout
  • Careers
  • Contact
PartnersCommunity Edition
Book a demoCommunity Edition

Meet the team at AWS re:Invent 2026

Las Vegas · 30 Nov – 3 Dec

Book time with Tracebit co-founder and CEO Andy Smith, or come and find the team at booth 1336 in The Venetian.

Book a meeting with Andy Find us at booth 1336

Book a meeting at re:Invent

Meet Andy in Las Vegas

Andy Smith

Andy Smith

Co-founder and CEO, Tracebit

  • A walk through our new AI-attacker research: which traps stalled which models, and why.
  • A live demo: canaries in your AWS accounts, and the alert your SOC gets when one is touched.
  • Your environment: where deception fits alongside what you already run, and the roadmap.

Open the booking page in a new tab

Booth 1336

Come by our booth to chat to the team

No meeting needed. Stop by booth 1336 to see Tracebit live, or have the team take you through our new research on what stops AI attackers.

Booth

1336

Venue
The Venetian Convention and Expo Center, Las Vegas
Dates
30 November – 3 December 2026

New research · Making attacking agents pay

Bankrupt AI attackers

Make them waste their tokens.

Canaries alert your SOC the moment an attacker touches them. Then they burn through the attacker's budget. We turned seven AI models loose on AWS environments in 773 assessments, seeded with nine Context Bomb techniques. Here is what happened to the attackers.

23×

more expensive per objective

$1.33 at baseline, $31.10 with deception, across six models with complete cost data.

91%

fewer objectives reached

2.48 per run at baseline, 0.23 with deception, across all seven models.

0/40

runs reached an objective

Kimi, Qwen, DeepSeek and GPT-5.6 Cyber under combined deception, down from 39 / 40.

7.8×

slower to first objective

Staged encryption against GLM 5.3: 4.0 minutes at baseline, 31.2 with deception.

Want the deeper dive? Ask our team at booth 1336, or book time with Andy and he'll walk you through it.

Read the research Book a deep dive
Soc 2 Type 2 imageCheckmark imageAWS Qualified software illustration
  • G2★4.9 / 5
  • Gartner Peer Insights★4.9 / 5
PLATFORM
AWS
Azure
CI/CD
Google Cloud
Identity
Kubernetes
Workstations
Credentials & artifacts
Connectors
USE CASES
AI Attack Detection
AI Agent Oversight
Cloud & Kubernetes Breach
Insider Threat Detection
Supply Chain & CI/CD Attack
Workstation Compromise
COMPANY
CustomersResearchAboutCareersContactFAQStatusCommunity EditionFree Enterprise Edition Trial
SOCIAL
© 2026 Tracebit
Privacy PolicyTerms of ServiceCookie Settings
Ask Tracebit

Loading chat…

Contact the team