Today, @Abnormal is launching a new AI Security product suite to help customers govern AI, secure AI, and defend against rogue AI. This new suite is built on the same behavioral AI technology that 30% of the Fortune 500 use for email and identity security. Abnormal designs, trains, and deploys our own AI, and we've spent nearly a decade mastering how to understand the normal behavior of business, detect behavioral anomalies, and respond at machine speed.  We're extending our behavioral modeling to include non-human identities, applications, and cloud identities. This enables the same behavioral AI to now power 5 new products in our AI Security product family: 1) AI Governance discovers your AI, uncovers shadow AI, shows who is using what, and how much it costs 2) AI Cloud Security detects + contains AI cloud breaches with behavioral AI and honeypots 3) AI Employee Guardrails enables employees to use AI without leaking data or triggering unintended actions 4) AI Agent Security inventories your agents and ensures they behave as intended 5) AI Security Workbench lets you use AI to run investigations with Abnormal models and without the guardrails AI Governance is already GA and can be installed in one click from our app store. AI Cloud Security is available today for priority customers, and the other products are in private preview for design partners. Learn more on our website and talk to your customer rep to cut the waitlist! abnormal.ai/blog/ai-security…

Last week, @Abnormal announced AI Cloud Security, and I wanted to share our internal prototype about how we’re using Abnormal’s core behavioral AI detection and response technology to stop the next generation of AI attacks. AI Cloud Security is built on the same core architecture and technology that we use for our email and identity security products that protect ~30% of the Fortune 500 today: 1) Map all the human, non-human, and cloud identities 2) Build behavioral models to understand baseline / expected behavior 3) Detect behavioral anomalies 4) Respond autonomously at machine speed We’ve spent the last several years building the core technologies, and I’m excited to start sharing our new AI security products as we build them with this audience. These products will be available to customers this year, but in the meantime, we are working with select design partners to ensure their success in the largest, most sophisticated enterprise environments. We are learning and building every day, and I’m excited to share more as we go!

It took 3 hours and one token to fully compromise an enterprise cloud environment. Anthropic's report from today yet another wake-up call. AI-powered attacks are already here, you just haven't heard about them yet.

Today's episode of Enterprise AI Innovators features Ogi Redzic, Chief Digital Officer of @CaterpillarInc, a 103-year-old company with a 3,000-person digital team and data flowing in from 1.6 million connected assets around the world. The company sits on both sides of the AI wave. It sells the prime power new data centers run on, and it runs AI across one of the largest machine fleets on the planet. A few takeaways from our conversation: 1) Caterpillar shipped its Cat AI Assistant in 10 months, a launch that size used to take years. It only moved that fast because the team spent years pulling every siloed app onto one platform, Helios. Customers ask a question now instead of learning a stack of apps. 2)About 39,000 technicians retire or leave the trade in the US every year. Caterpillar's support system covers 50,000 documents and equipment going back 56 years, so a first-year tech can stand at a machine, ask how to replace the hydraulic pump, and get the answer on a tablet. 3) Caterpillar has a head start on autonomy. Its mining trucks have driven hundreds of millions of kilometers with no reported safety incidents. Construction is harder, and Ogi's next bet is orchestrating a whole job site so the work runs straight through. Full episode below: enterprisesoftware.blog/podc…

Eight years ago we bet that behavioral AI, modeling what normal looks like for every identity, was the right way to stop attacks that signatures never could. We started with email because that's where the pain was. The future came at us faster than any of us expected. The Hugging Face incident made it concrete. Highly capable agents chained ordinary flaws and gaps into a path to production infrastructure. Our customers asked for the same engine in their cloud, paired with the strongest defensive models available. Today, @Abnormal is announcing AI Cloud Security, extending our behavioral AI engine to the cloud to detect risky or malicious AI agent behavior inside customer environments, with @OpenAI models supporting investigation and response. AI Cloud Security is now in private preview for select Abnormal AI customers, and Abnormal is now part of OpenAI's Daybreak Defense Network alongside the OpenAI Cyber Summit. Check out @sanjays_tweets blog to learn more: abnormal.ai/blog/openai-rogu… P.S. Yesterday I was on a webinar walking customers through an early demo that Sanjay and I have been building recently. This moment feels a lot like the early days of Abnormal, a direct pipeline from customer feedback to product building, and I'm excited for what's next!

I feel proud and excited to partner with OpenAi to help protect and secure the world in this AI era!

How do you catch an attacker sending from a real account? Real domain, real history, real relationships. Every signal your defenses were built to trust says "friend." The only question left is whether the account is behaving like itself. Two favorite details from how our team cracked this: You can't retrain on misses you can't find. So they mined the "I was hacked, please ignore my last email!" apologies victims send afterward, then traced back 72 hours to find the attack. The victim labels the attack for you. Adding email content made detection worse. The signal was sender behavior measured against that account's own baseline. Behavior is the tell. In the last 30 days this model caught ~9,000 campaigns that no other detector flagged. This is what building AI to fight AI actually looks like.

This week, @Abnormal announced a partnership with @OpenAI to accelerate secure enterprise AI adoption and strengthen cyber defense. At Abnormal, we train and deploy Behavioral AI to stop highly sophisticated cybersecurity attacks that conventional tools miss, and OpenAI builds the frontier models that help defenders investigate those threats faster. As generative AI gives attackers new speed and sophistication, defenders need both kinds of innovation working together to keep up. We're also joining the OpenAI Daybreak Cyber Partner Program as an early security partner, and we're exploring how OpenAI's frontier models and Codex Security can complement our Behavioral AI, including new capabilities to help security teams monitor AI agent activity and catch anomalous behavior. Abnormal is expanding our own use of OpenAI internally, across our business operations and software development. We're early on all of this, and there's a lot more coming: abnormal.ai/blog/abnormal-op…

Why does agent infrastructure matter for stopping crime with AI? Because attackers don’t retry politely. They attack 24/7 at machine speed, and your defense can’t drop an investigation because one API call failed. Our agents run tens of thousands of security investigations a day. Each one makes dozens of model + service calls. At that scale, “mostly reliable” means missed attacks. So we rebuilt our platform on durable execution: every investigation survives failures, resumes where it left off, and finishes the job. Deeper investigations → catch attacks that signals miss. Reliable execution → catch them every time. You can’t stop crime with AI if your AI gives up halfway.