Fabraix
fabraix.com
Actively hiring
Vulnerability Management
Penetration Testing
Cybersecurity AI
Follow
Overview
News
Technologies
Salaries
Products
People
Growth
Offices
Financials
Overview
Fabraix builds red-teaming AI agents that continuously find security vulnerabilities in customer-facing AI.
News
Bounding the Blast Radius: A Survey of Prompt-Injection Defenses for LLM Agents
Prompt injection has no known general solution. We organize the defense landscape into a four-layer taxonomy, analyze the documented failure mode of each layer, and argue for composing defenses under explicit cost and latency budgets, then verifying the r
Read more
Report
Adversarial Cost to Exploit (ACE): A Dynamic Benchmark for AI Agent Security
We introduce Adversarial Cost to Exploit (ACE), measuring the token expenditure an adversary must invest to breach an LLM agent. Six budget-tier models show an order-of-magnitude variance in adversarial cost.
Read more
Report
2026 so far has matched all of 2024 for AI security incidents.
We catalogued 41 AI security incidents since January 2024. Q1 2026 alone matched all of 2024. March had 8 - the worst single month on record.
Read more
Report
SWE-bench will hit 90% this year.
Six models are within 1.3 points on SWE-bench Verified. Binary benchmarks are saturating - what will replace them?
Read more
Report