Real Alpha
Watch modeSubscribed

All-In · 3 min · Oct 21, 2026

AI vulnerability findings create an asymmetric race

Nikesh Arora

Chairman & CEO, Palo Alto Networks

Chairman and CEO of Palo Alto Networks, which he grew into a $300 billion cybersecurity leader.

  • Chamath PalihapitiyaHost · Founder & CEO, Social Capital

    Led Facebook's growth team; his firm backed Slack and AI chipmaker Groq.

  • Jason CalacanisHost · Founder & CEO, LAUNCH

    Early Uber angel; sold Weblogs Inc. to AOL; hosts This Week in Startups.

  • David SacksHost · Co-founder & partner, Craft Ventures

    PayPal's first COO, founded Yammer; White House AI & crypto czar 2025–26.

  • David FriedbergHost · CEO, Ohalo

    Founded The Climate Corporation, sold to Monsanto for about $1.1B in 2013.

About this segment

Nikesh says current models surface many real flaws, but attackers need only one while defenders must verify and patch each finding. That gap gives AI-assisted discovery different value on offense and defense.

From “Arvind Krishna & Nikesh Arora: Quantum Threat, Open vs Frontier AI, AI Cyber Warfare & SaaS Collapse”, Oct 21, 2026. Watch the full episode

“they just need one vulnerability they don't need 10 vulnerabilities”
Nikesh Aroraat 17:05 · All-In17:05

Takeaways

  1. Nikesh estimates 60–80% of reported findings were real.
  2. Attackers need one exploitable flaw; defenders must clear them all.
  3. AI findings still need testing before teams trust a patch.
Watch 2:34 from 15:25

In this episode

Full episode
  1. 0:28Quantum-safe security needs to start before quantum computers arrive6:20
  2. 9:01Open-weight models could handle most enterprise AI at lower cost6:21
  3. 35:19AI could consolidate hundreds of enterprise software vendors4:21

  4. 15:25AI vulnerability findings create an asymmetric race2:34

  5. 30:04AI application stacks can make user context a moat4:17Up next
  6. 18:03AI-written code can outperform when agents run evals3:09

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

  1. 15:29

    When you're sitting with these guys, is part of you thinking they're looking at my business and they're like, wow, this is a great business? It's like, do you think that they're... And what is the actual technical boundaries that you expect to be respected? And the reason I ask this is, we've talked to a couple of other CEOs. I think Mark Benioff does the best example of this, where we've talked about ZDR, and he's been incredibly aggressive about making sure there's zero data retention and doing that on behalf of his customers.

  2. 16:00

    What's so interesting about these EDR and all of this stuff is like none of the models say that they won't generate synthetic data. It's literally not in there. You guys should look. There's all these outs. There's all these trap doors. So I'm just wondering like when you're in the room, what are you thinking? When you're in the room, you come out and say, these guys are not going to get into my business. They're not going to be able to get into my business. Okay. Because if you think about when Mythos came out, right, we've been paying billions of dollars for encoding agents. Well, guess what? We're fundamentally, not everybody's got ZD out on coding. There's tons of people out there using cloud code. We have no issue about that data training the model. So we are training these models with end users doing coding. These models

  3. 16:46

    are getting better and better doing coding. Well, guess what? If you teach a model what good code looks like it also can figure out what bad code looks like so the person that missiles did it figured out what bad code looks like and if you let it loose against your code it'll tell you what is bad now what's fascinating is when i did that even then it was fine it was hallucinating 30 of the time right but the problem is the reason it's perfect for attackers they just need one vulnerability they don't need 10 vulnerabilities if you're a defender you need to find every vulnerability and patch it so all the people in zoom i think have been through this journey in the last 12 months is that everybody here has gone, tested them, either with Mythos or with some cousin of Mythos or some open source models and discovered every vulnerability

  4. 17:29

    has to then be tested, understood, you know, figured out. And I would guess in this room, if I show hands, people say that numbers roughly 60 to 80% have been real and 20% have been hallucinations. So from that perspective, the models are great to be able to tell you what the vulnerability But I don't even think anybody in this room, if you ask them to raise their hand, will tell you that they're not going to even use Claude Code to patch the vulnerability without having to test it again.