Itamar Tal

VP engineering @guardicore / Akamai. Distributed high-performance security systems. Linux, opensource, virtualization, networking enthusiastic

[𝙿𝚊𝚟𝚎𝚕 𝙶𝚞𝚛𝚟𝚒𝚌𝚑 @PashaGur]

Finding vulnerabilities was always just the first step for @Tenzai_Labs . Rapidly reducing the exposure window is what really matters.
Particularly exxcited to announce our collaboration with @Akamai's WAF to fix the web vulnerabilities we find.

[Tenzai @Tenzai_Labs]

Here we go! 🚀

Most enterprises are still running security programs built for a world where software shipped quarterly and adversaries moved slowly. Neither is true anymore.

That's why @Ahead teamed up with @Tenzai_Labs - strengthening cyber resilience for enterprises, everywhere.

Read: https://t.co/R6eJEHDI2v

[Tenzai @Tenzai_Labs]

Plot twist!

Tenzai's AI hacker now files the ticket, assigns the fix, re-runs the exploit, and closes the loop.

We tried to keep it focused on breaking things. It had other ideas.

Introducing: Automation Rules!

Read it, watch the demo, take it for a spin:

https://t.co/nGHspA70DU

[Tenzai @Tenzai_Labs]

AI is moving fast, they say...

Well: *For the first time* an autonomous AI hacker doesn’t just identify and validate exploitable risk; it deploys immediate WAF rule mitigations and guides permanent remediation, closing the loop from discovery to defense at machine speed, through leading security platforms, including @Akamai.

This is Tenzai --> https://t.co/yQGIvsYOkq

[Tenzai @Tenzai_Labs]

PSA: A vulnerability introduced on Monday shouldn't wait until Saturday night.

That's why we built incremental testing: pentesting that runs automatically after every deployment, scoped to what changed. → https://t.co/WQe2vHK7xa

Thank us later 🫣

[𝙿𝚊𝚟𝚎𝚕 𝙶𝚞𝚛𝚟𝚒𝚌𝚑 @PashaGur]

Development velocity keeps increasing, but budgets need to stay flat.
So we built an ability to only actively test and review code parts that changed. This helps reduce token cost and time.

[Tenzai @Tenzai_Labs]

(This is kinda cool, if we do say so ourselves.)

We tried to keep Tenzai focused on breaking things. It had other ideas.

Introducing: Automation Rules!

See it in action:

[𝙿𝚊𝚟𝚎𝚕 𝙶𝚞𝚛𝚟𝚒𝚌𝚑 @PashaGur]

Soon, malicious hacking agents will be relentlessly trying to break into applications and networks.
They wont reuse generic vulnerabilities or run off-the-shelf tools. They will look for specific weaknesses in specific apps, networks, configurations, identities, and environments.
The future can't be about protecting generic networks from generic attacks.
The future will be about tailored protection defending specific networks from specific vulnerabilities.
That's why we need to continuously simulate intelligent attacks paths, understand what vulnerabilities allowed for it, and instantly translate the findings into better protections, detections, and behavioral signals.
We need a constant purple team between offensive agents and defensive agents.
https://t.co/mgN6y1dq4F

[𝙿𝚊𝚟𝚎𝚕 𝙶𝚞𝚛𝚟𝚒𝚌𝚑 @PashaGur]

As promised a more detailed view into GLM 5.2s security capabilities and associated costs compared to the "opentropic" models.

[Tenzai @Tenzai_Labs]

Stuff our research team is working on: https://t.co/VGMyLVIyAp

[𝙿𝚊𝚟𝚎𝚕 𝙶𝚞𝚛𝚟𝚒𝚌𝚑 @PashaGur]

H1 Q2 26 VDP Biz US #1
~$20k >90 days

https://t.co/aZcRTRfRMO

[𝙿𝚊𝚟𝚎𝚕 𝙶𝚞𝚛𝚟𝚒𝚌𝚑 @PashaGur]

Need to quickly test your apps with Tenzai?

Now you can quickly get it through your existing AWS account, use your commit, custom pricing via private offer.
Check it out ➡️
https://t.co/VfnD1xupNK…

[Tenzai @Tenzai_Labs]

We let our AI hacker loose on hundreds of thousands of human hackers and landed in the top 1%.

*Now it's loose in the @awscloud (AWS) Marketplace* 🚀

Yep, we're live on AWS! Autonomous, enterprise-grade pentesting, across every application you ship, continuously.

Get it through your existing AWS account, use your commit, custom pricing via private offer. No red team to schedule. Check it out ➡️
https://t.co/tGicYTcI1O

[Tenzai @Tenzai_Labs]

Woah 🧐 -->

The entry point was open user registration. Anyone could sign up. The exit was full control of a production database server.

The steps in between crossed six (🤯) different domains, the kind of chain most security tools never see because each one only looks at its own piece.

In our latest Tenzai in the Trenches by @ace__pace , our AI hacker found some pretty crazy stuff in a multinational enterprise environment. Read on -->
https://t.co/LrC7FLHCkp

[Tenzai @Tenzai_Labs]

New drop ⬇️
Tenzai now tests AI applications.

The world's most powerful AI hacker just expanded its scope. Not as a separate module, but as the next layer of the same attack surface we've been breaking for web apps and APIs.

Read it → https://t.co/DB8R0hngrw

[Tenzai @Tenzai_Labs]

Tenzai has been named to @Redpoint’s 2026 InfraRed 100, recognizing the companies shaping the future of infrastructure and AI. Congratulations to all the companies featured this year! ⬇️

[Tenzai @Tenzai_Labs]

We're looking for (ambitious!) humans obsessed with working on problems that matter and making an impact quickly, as @Tenzai_Labs grows rapidly within an emerging category of security! If you thrive building something from zero and helping a new idea take hold in the world, keep reading --> https://t.co/Z7J34EpPrr

[Tenzai @Tenzai_Labs]

How do you get to the top 1%? Roy Miara, an (OG!) member of the technical staff at @Tenzai_Labs, shares his thoughts on building an AI hacker from scratch that can beat 99% of humans: (Full read 👉 https://t.co/6aqopP1DvT)

[Tenzai @Tenzai_Labs]

Let's go! @PashaGur @itamartal @OfriZiv @ArieZeitlin @anermazursky @NathanPerdue2 💪 https://t.co/66mLE6V0qm

[Tenzai @Tenzai_Labs]

Humans simply don't scale. So we created AI hackers that find vulnerabilities created by humans or other AI agents - and we're seeing they're better than humans 99% of the time 🎯 Thanks @NYSE for a riveting conversation with CEO @PashaGur on the sidelines of @OneRSAC.