1. X
  2. XBOW
Log inSign up
XBOW
364 posts
Image
user avatar
XBOW
@Xbow
The autonomous offensive security company redefining cyber defense for the AI era.
Seattle, Washington, USA
xbow.com
Joined May 2007
13
Following
11.8K
Followers
RepliesRepliesMediaMedia

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
  • user avatar
    XBOW
    @Xbow
    2h
    The '90s brought the internet and built the playground we've been hacking ever since. Come throw it way back with us at Black Hat in Las Vegas! 📍Booth #3448 🔗xbow.com/events/black-h…
    Image
    00:00
    512
  • user avatar
    XBOW
    @Xbow
    23h
    What’s your preferred model for bug bounties? (If you use a mix, comment in thread)

    When you make a selection it cannot be changed

    108 votes2m left
    1.7K
  • user avatar
    XBOW
    @Xbow
    Jul 10
    Our AI research team has been analyzing affordable AI models as they are released, most recently Grok 4.5. Here’s what we found: Grok 4.5 performs exceptionally well when you have more than a shoestring budget but still care about cost. It does not win at every price point, but
    29K
    user avatar
    XBOW
    @Xbow
    Jul 10
    Replying to @Xbow
    Image
    4.5K
    user avatar
    XBOW
    @Xbow
    Jul 10
    Read the full research here:
    xbow.com
    Grok 4.5 for Offensive Security: Performance vs. Cost | XBOW
    XBOW evaluates Grok 4.5 against GPT-5.5, Mythos, and GLM, showing where it delivers the best balance of cost and offensive security performance.
    73K
  • user avatar
    XBOW
    @Xbow
    Jul 10
    The rise of affordable AI models is changing the economics of offensive security. Muse Spark 1.1 and GLM aren’t the best models we’ve tested for finding vulnerabilities. That’s not the point. A thread on why we think "best" is the wrong question. 🧵
    13K
    user avatar
    XBOW
    @Xbow
    Jul 10
    Replying to @Xbow
    The future of AI-enabled offense isn't about the most powerful model. It's about what happens when good-enough gets cheap enough to run everywhere. That's the part defenders need to take seriously.
    895
    user avatar
    XBOW
    @Xbow
    Jul 10
    What do you think? Read our details thoughts + research here:
    xbow.com
    Muse Spark vs. GLM: AI Models for Offensive Security | XBOW
    XBOW compares Muse Spark 1.1, GLM, Mythos, and GPT-5.5 to evaluate how cost, capability, and performance are reshaping AI-powered offensive security.
    897
  • user avatar
    XBOW
    @Xbow
    Jul 9
    The 80s, 90s, and early 00s were a crazy time in computing where things moved at a rapid clip and technology still had _style_ 🤭 ... can we go back and hack all of it? If you're in NYC attending @SummerC0n this weekend, @moyix + @kallsyms will be sharing their process for
    Image
    3.6K
  • user avatar
    XBOW
    @Xbow
    Jul 8
    POV: Ctrl+C, Ctrl+V but it didn't quite work.
    Image
    2.8K
  • user avatar
    XBOW
    @Xbow
    Jul 6
    In light of the recent debate surrounding access to Claude Fable 5 & Mythos 5, should companies use open-weight AI models like GLM-5.2/Kimi?
    ✅ Yes, ROI justifies it61.8%
    ✔️Only with guardrails3.6%
    ❌ No, risks are too high9.1%
    💭 Still evaluating25.5%
    55 votesFinal results
    1.6K
  • XBOW reposted
    user avatar
    Tierra de Hackers
    @tierradehackers
    Jul 6
    Hablo de Mythos con @fede_k , líder del Security Lab de @Xbow . ¿Qué hay de cierto en eso de que "hackeó a la NSA"?, por qué dice que "no es que él sea bueno, es que los sistemas son malos", y qué cambia de verdad con la IA en la ciberseguridad.
    Image
    Experto en IA ofensiva: ¿Una IA hackeó a la NSA? ¿Qué usaban Windows XP?
    From youtube.com
    4.6K
  • user avatar
    XBOW
    @Xbow
    Jul 2
    Developing a multi-stage attack chain manually can take anywhere from days to months. Working with XBOW, Moderna compressed that timeline to less than a day. Read full success story: na2.hubs.ly/H06kRRr0
    Image
    1.6K
  • XBOW reposted
    user avatar
    Brendan Dolan-Gavitt
    @moyix
    Jul 2
    These will be very fun when we get to talk about them (V8)
    Image
    5.3K
  • user avatar
    XBOW
    @Xbow
    Jul 1
    10 key questions to ask before building an AI-powered offensive security tool internally? See our new brief: na2.hubs.ly/H06kScv0
    Image
    1K
  • user avatar
    XBOW
    @Xbow
    Jun 30
    Most security tools answer one question: what weaknesses exist here? They return a list, each item scored on its own.  Real attackers do not work this way.  They compose.  They take a low-severity foothold, use it to unlock the next, and chain primitives together until
    Image
    1.1K
Advertisement
Advertisement