Log inSign up
user avatar

Andrea Michi

depthfirst
@andreamichi
CTO @depthfirstlabs - Building a modern Security Operating System. Prev RL post-training Gemini @GoogleDeepMind
San Francisco
Joined March 2009
1,351
Following
2,801
Followers
RepliesMedia
  • user avatar
    Andrea Michi
    depthfirst
    @andreamichi
    Aug 19
    Scaling defensive cybersecurity is top of mind. Today we launched dfbench, a held-out cyber benchmark built by our security research team. We focus on detection, validation and differential analysis. We’ll keep dfbench up to date as cyber capabilities continue to evolve
    user avatar
    depthfirst
    @depthfirstlabs
    Aug 19
    Today we are announcing dfbench, a cybersecurity benchmark we developed to evaluate frontier models and agentic systems. A few weeks ago we posted about how it is time to graduate from CyberGym Level 1 as the primary measure of cyber capability, in favor of newer benchmarks.
    Image
  • user avatar
    Andrea Michi
    depthfirst
    @andreamichi
    Aug 13
    This was a really fun human–agent collaboration challenge. We want to keep your threat model up to date while also giving you the flexibility to tune it as needed. The threat model will then be used across all our products, from Code and Security Reviewer to Agentic Pentester.
    user avatar
    depthfirst
    @depthfirstlabs
    Aug 13
    Introducing Threat Model: living security artifacts that capture each repository’s architecture, trust boundaries, attacker-controlled inputs, and assumptions so vulnerability discovery reflects how your application actually works. Most security tools start every finding from
    Image
  • user avatar
    Andrea Michi
    depthfirst
    @andreamichi
    Aug 8
    Yep. We often ignore that profitability will be shaped by underlying power laws. Cybersecurity is a good example: there’s an economic frontier. You’re only as secure as the cost of attacking you, so every incremental improvement matters.
    user avatar
    Naval
    @naval
    Aug 8
    Open-source AI models do not threaten frontier lab profitability. The most valuable sectors of the economy - investing, product development, war, cybersecurity, even scientific discovery - are adversarial and competitive in nature. You pay to win, or someone else will.
  • user avatar
    Andrea Michi
    depthfirst
    @andreamichi
    Jul 30
    Today we're announcing dfs-large1, our newest cybersecurity model that achieves best-in-class performance on vulnerability detection tasks. Besides frontier AI labs, only a handful of companies have built specialized models that reach the state of the art in their domain. We're
    Image