AIR

AIR secures every layer of the agentic stack, so enterprises can adopt AI safely.

[AIR @air__security]

𝗔𝗜𝗥 𝗶𝘀 𝗡𝗼𝘄 𝗔𝘃𝗮𝗶𝗹𝗮𝗯𝗹𝗲 𝗼𝗻 @googlecloud 𝗠𝗮𝗿𝗸𝗲𝘁𝗽𝗹𝗮𝗰𝗲!

Google Cloud customers can now procure AIR through their existing account - consolidated billing, and drawing down on existing committed spend.

“Bringing AIR to Google Cloud Marketplace will help customers quickly deploy, manage, and grow the company’s AI supply chain security platform on Google Cloud’s trusted, global infrastructure” Dai Vu, Managing Director, Marketplace & ISV GTM Programs at Google Cloud. “AIR can now securely scale and support enterprises using its platform to help secure their AI agent fleets.”

[Yair Saban @ItsYairSaban]

Today, @air__security comes out of stealth, and I couldn’t be more excited.

Our goal is simple: make sure your next prompt is secure.

I’m incredibly grateful to our team and investors @BogieBalkansky @DeanMeyerrr @DavidCahn6 @pat_backhouse @CaspiOmri @Weiner_Lia

Let’s go!

[AIR @air__security]

@Tecnology @meirorbach ❤️

[Bogomil Balkansky @BogieBalkansky]

We at @sequoia believe the @air__security team was the strongest to come out of unit 8200 this year, and we are delighted to be in business with them.
Little attention has been paid to the supply chain aspect of AI security, and this is what Air is fixing.

[AIR @air__security]

https://t.co/rDEbQ6UDkX

[AIR @air__security]

WE'RE ON AIR!

𝗔𝗜𝗥 𝗶𝘀 𝗼𝗳𝗳𝗶𝗰𝗶𝗮𝗹𝗹𝘆 𝗼𝘂𝘁 𝗼𝗳 𝘀𝘁𝗲𝗮𝗹𝘁𝗵 - 𝘄𝗶𝘁𝗵 $𝟱𝟬𝗠 𝗶𝗻 𝗳𝘂𝗻𝗱𝗶𝗻𝗴 𝗹𝗲𝗱 𝗯𝘆 @sequoia , and @Greenoaks .

We started AIR with a simple belief: as AI agents become more capable, the security model around them has to evolve.
Agents are no longer just generating answers. They’re connecting to tools, skills, MCPs, plugins, websites, and internal data - and taking real actions across the enterprise.

Today, we’re introducing the 𝗰𝗼𝗻𝘁𝗲𝘅𝘁 𝗳𝗶𝗿𝗲𝘄𝗮𝗹𝗹 𝗳𝗼𝗿 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀: a new security layer built to continuously analyze what enters an agent’s context, control what can run, and stop threats in real time before they turn into action.

We’re incredibly grateful to our investors, partners, customers, and everyone who believed in us early.

This is a huge milestone for our team, and an even bigger beginning ❤️

Read our full story in the first comment 👇

[AIR @air__security]

🐝 OWASP Top 10 for Skills Standard is now publicly available!

Earlier this year, Niv Hoffman, our CTO & Co-founder, joined Ken Huang, Bhavya Gupta, and the rest of the @owasp leadership team to build the security standard for the skills ecosystem.

The new standard is a must-read for any security practitioner looking to secure this new software layer and the supply chain risks it introduces.

Looking forward to implementing it together with the foundation, the labs, and the whole open source community!

Check it out: https://t.co/5MSwGLUy2O

[AIR @air__security]

🎙️ Niv Hoffman on AWS Israel's podcast!

Niv, Our CTO & Co-Founder, joined Gili Nachum from @awscloud on AWS Israel’s “Building AI” podcast to dive into the growing skills ecosystem - the risks hiding behind seemingly useful AI add-ons, and how everyone can use them more safely.

In the episode:

▪️ 𝗛𝗼𝘄 𝗮 𝗳𝗮𝗸𝗲 𝘀𝗸𝗶𝗹𝗹 𝗲𝘅𝗽𝗼𝘀𝗲𝗱 𝟮𝟲,𝟬𝟬𝟬+ 𝗔𝗜 𝗮𝗴𝗲𝗻𝘁𝘀 - and what the experiment revealed about how easily trust can be exploited.

▪️ 𝗛𝗼𝘄 𝗦𝗸𝗶𝗹𝗹𝗝𝗮𝗰𝗸𝗶𝗻𝗴 𝘄𝗼𝗿𝗸𝘀 - and why determining whether a skill is safe is much harder than it seems.

▪️ 𝗛𝗼𝘄 𝘁𝗼 𝘂𝘀𝗲 𝘀𝗸𝗶𝗹𝗹𝘀 𝗺𝗼𝗿𝗲 𝘀𝗮𝗳𝗲𝗹𝘆 - practical steps for evaluating skills, protecting development environments, and building a secure internal marketplace.

👉 Watch the full episode - https://t.co/kTSGifwN9o

[AIR @air__security]

Huge thanks to Niv Hoffman, Ken Huang, Omar A. Turner, and Apostol Vassilev for a great discussion on the OWASP Top 10 for Agentic Skills Initiative at Black Hat.

And thanks to everyone who joined us!

[AIR @air__security]

𝗕𝗹𝗮𝗰𝗸 𝗛𝗮𝘁 𝟮𝟬𝟮𝟲… 𝘄𝗵𝗮𝘁 𝗮 𝗿𝗶𝗱𝗲! 🎪

Thank you to everyone who stopped by AIR Booth, spun the wheel, met the stars of the circus, and shared thoughts and challenges
with our team.

Until next time 💫

[AIR @air__security]

Launching the AIR website.

The first step in defining supply chain security for AI agent fleets. It’s a big moment for us.

This is where our story begins, but it’s only the beginning.

Explore the new website: https://t.co/WzZc7RCrRA

[AIR @air__security]

👀👀👀

[AIR @air__security]

https://t.co/WPW8pZf0xS

[AIR @air__security]

🚨 BREAKING

Our latest research uncovered multiple AI skills impersonating trusted AI brands like Anthropic and OpenAI - and passing public security scanners.

How?

🚩 They verified the package, not the repository behind it.
🚩 They trusted the dependency name, not who actually owned it.
🚩 They validated what the skill declared - not whether those claims were true.

As AI agents increasingly rely on external packages, repositories, and third-party resources, a one-time scan is no longer enough.

Trust has to be continuously verified.

That’s exactly why we built AI Add-on ScannAIR - to continuously vet every dependency, repository, package, and publisher behind every AI add-on, instead of trusting what it claims to be. 🎭
https://t.co/Z53zYHVuyS

The full research is in the first comment 👇

[AIR @air__security]

LET'S GOOOOO

[AIR @air__security]

🚨 NEW RESEARCH

𝗔 𝘁𝗿𝘂𝘀𝘁𝗲𝗱 𝗔𝗜 𝘀𝗸𝗶𝗹𝗹 𝗰𝗮𝗻 𝗯𝗲𝗰𝗼𝗺𝗲 𝗺𝗮𝗹𝗶𝗰𝗶𝗼𝘂𝘀 𝘄𝗶𝘁𝗵𝗼𝘂𝘁 𝗰𝗵𝗮𝗻𝗴𝗶𝗻𝗴 𝗮 𝘀𝗶𝗻𝗴𝗹𝗲 𝗹𝗶𝗻𝗲 𝗼𝗳 𝗰𝗼𝗱𝗲.

In our latest research, we uncovered 𝗦𝗸𝗶𝗹𝗹𝗝𝗮𝗰𝗸𝗶𝗻𝗴 - a new supply chain attack that allowed us to hijack 925 AI skills, affecting 134K+ AI agents.

𝗧𝗵𝗲 𝗮𝘁𝘁𝗮𝗰𝗸 𝘄𝗲𝗻𝘁 𝘂𝗻𝗻𝗼𝘁𝗶𝗰𝗲𝗱 𝗯𝘆 𝗮𝗹𝗹 𝗽𝘂𝗯𝗹𝗶𝗰 𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝘀𝗰𝗮𝗻𝗻𝗲𝗿𝘀.

🚩 The takeaway is simple: Don’t just vet the skill. Vet everything it depends on. Continuously.

Read the research: https://t.co/cSU58K8WKN

[AIR @air__security]

𝟭𝟳,𝟴𝟮𝟮 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲 𝘀𝗸𝗶𝗹𝗹𝘀. 𝟲.𝟳𝗠+ 𝗶𝗻𝘀𝘁𝗮𝗹𝗹𝗮𝘁𝗶𝗼𝗻𝘀. 𝗢𝗻𝗲 𝗺𝗮𝘀𝘀𝗶𝘃𝗲 𝗮𝘁𝘁𝗮𝗰𝗸 𝘀𝘂𝗿𝗳𝗮𝗰𝗲.

🎪Welcome to the Circus of Skills

Many of AI add-ons rely on external instruction sources that can change at any moment - turning trusted add-ons into an enterprise risk.

𝗜𝗻 𝘁𝗵𝗶𝘀 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵, 𝘄𝗲 𝗿𝗲𝘃𝗲𝗮𝗹:
🤡 17,822 open-source skills vulnerable to Untrusted External Instructions Sources
🤡 Why this attack vector now appears in the OWASP Agentic Skills Top 10 Risks
🤡 What enterprises need to do before trusting AI add-ons at scale

Read he full research: https://t.co/Dmc5v7oK55

[AIR @air__security]

𝙊𝙣𝙚 𝗺𝗮𝗹𝗶𝗰𝗶𝗼𝘂𝘀 𝙨𝙠𝙞𝙡𝙡 𝙨𝙚𝙞𝙯𝙚𝙙 𝙛𝙪𝙡𝙡 𝙘𝙤𝙣𝙩𝙧𝙤𝙡 𝙤𝙛 𝙤𝙫𝙚𝙧 𝟮𝟲,𝟬𝟬𝟬+ 𝘼𝙄 𝙖𝙜𝙚𝙣𝙩𝙨 - 𝙖𝙣𝙙 𝙣𝙤𝙩 𝗮 𝘀𝗶𝗻𝗴𝗹𝗲 𝙨𝙚𝙘𝙪𝙧𝙞𝙩𝙮 𝙨𝙘𝙖𝙣𝙣𝙚𝙧 𝗱𝗲𝘁𝗲𝗰𝘁𝗲𝗱 𝗶𝘁

AI skills are changing how enterprises adopt AI. They’re discovered on social media, shared on GitHub, and installed into AI agents in seconds. Security, however, hasn’t caught up.

𝗜𝗻 𝘁𝗵𝗶𝘀 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵, 𝘄𝗲 𝗿𝗲𝘃𝗲𝗮𝗹:
🚨 How the attack worked
🚨 Why existing scanners failed to detect it
🚨 What enterprises must do to secure AI add-ons before they become the next major attack surface.

Read the full research: https://t.co/qV9tOJhwzo