Abnormal AI

4:30 PM | Off the Clock Happy Hour Together with @felicis, @AbnormalAI, and @1Password, enjoy great food, drinks, and high-signal conversations with security leaders. 👉 Register: bit.ly/4tKvhZN

Stop Chasing AGI And Start Fixing What’s Broken hubs.li/Q047NdN_0 Written by Mike Britton of @abnormalai

Email is central to the White House cyber EO as phishing drives most attacks, pushing a shift to AI-powered, proactive defenses to stop threats before they breach networks, says @AbnormalAI's Yejin Jang in this op-ed. #cybersecurity #CISO #infosec bit.ly/4slLpQu

Attackers aren’t breaking in anymore. They’re blending in. Today, attackers are using GenAI to deliver highly personalized attacks that are easy to deploy. As a result, defenders need to assume every attack is novel. And, if that's the case, static rules and legacy threat intelligence will always be one step behind. To stop an attacker who looks exactly like a trusted colleague, you have to stop looking for "bad" and start understanding "normal." Today, we’re introducing Attune 1.0—Abnormal’s behavioral foundation model. Attune doesn’t just block known threats; it learns the unique DNA of your organization across identity, context, and content. By understanding the baseline of human behavior, it can pinpoint the subtle anomalies that AI-driven attacks try to hide. Attune already powers 85% of detections across the Abnormal platform is delivering meaningful improvements in efficacy: roughly 150,000 more attack campaigns caught per week, with 50% higher precision than earlier systems. The era of "detecting bad" is over. The era of "knowing normal" has arrived. Learn more about Attune, plus other innovations across Cloud Email Security and Human Risk Management, via our CEO @evanreiser abnormal.ai/blog/attune-em…

We're officially one week away from RSAC 2026! Join us at our AI Hub at the Golden Eye Social for a week of innovation, networking, and a firsthand look at how we’re protecting the modern enterprise from the most sophisticated human-centric attacks. Here’s what’s happening at the AI Hub: ☕ Grab a coffee for our morning demos 🍸 Join us for afternoon cocktails 🤝 Book a 1:1 meeting with our experts 👇 Secure your spot here: abnormal.ai/rsac-2026-meet…

To our customers, this one's for you ✨ Kaskade is ready for Abnormal AI After Dark and we're counting down the days. Don't forget to register for our exclusive customer appreciation party at The Warfield, Tuesday March 24. Secure your spot—space is limited! abnormal.ai/rsac-party?utm…

Abnormal has been named to America’s Best Startup Employers 2026 by @Forbes! This recognition reflects the culture we’re building — high performance, mission-driven, and focused on stopping advanced email and AI-powered threats. As cybercriminals rapidly adopt AI, our team continues to push the boundaries of what’s possible with AI-native security—protecting global enterprises with precision and scale. Thank you to our employees for shaping the culture that makes this achievement possible. See the full list here: forbes.com/lists/americas…

Why your AI "Safety Instructions" are failing. When we talk about securing AI, we often mistake intent for control. Most organizations are securing their AI agents with what our CEO, @evanreiser, calls "polite wishes". You tell the agent what you want it to do, but instructions don't actually constrain what it can do. The result? A new, high-velocity failure mode: an agent that follows your instructions perfectly yet still executes the wrong action at machine speed. As we move toward an autonomous SOC, we have to stop treating AI security as a prompt engineering problem and start treating it as a behavioral one. To truly enable the secure use of AI, we must shift from the old paradigm to the new paradigm. The future of productivity depends on AI agents. But that future is only possible if we replace polite wishes with reliable, AI-native behavioral controls. Read Evan’s full breakdown on why the industry needs a paradigm shift in AI security: x.com/evanreiser/sta…

We are incredibly proud to announce that Abnormal AI has been named to The Agentic List 2026 🏆 As organizations adopt AI agents and automate critical workflows, the attack surface is evolving fast. At Abnormal, securing the Agentic Era isn’t just about protecting AI-driven enterprises—it’s about how we use AI and agentic AI ourselves to power that defense. Our AI-native Abnormal Behavior Platform leverages advanced AI and automation to understand identity, intent, and behavior across every interaction. By applying AI to continuously analyze signals, adapt in real time, and autonomously stop sophisticated attacks, we deliver protection that scales with the modern enterprise. We’re honored to be recognized for both how we secure AI-powered organizations and how we harness AI to do it. Thank you to our customers and partners who trust us to protect every identity across the F500 and beyond. agentconference.com/agenticlist/20…

One company. One focus. One mission: protect humans from cybercrime. SKO in Las Vegas was more than a kickoff. It brought our global team together to align on what we are building and why it matters. When you see the roadmap, the innovation, and the scale of what’s ahead, belief turns into urgency. We heard from the teams building and delivering our platform, and from the customers who depend on Abnormal to protect their people in a rapidly evolving threat landscape. That shared perspective matters. It sharpens execution, strengthens accountability, and connects our daily work to real-world impact. We are building the platform for protecting humans in the AI era. When the entire company moves with clarity and conviction, the impact accelerates.

Abnormal AI is excited to welcome Stephen Harrison as our VP of Product, Next Gen Products! 🎉 Stephen brings a rare combination of offensive security expertise, Fortune 500 CISO leadership, and hands-on product ownership. He understands how modern attacks unfold — and how security products need to evolve to keep pace. And his arrival comes at an important moment for Abnormal. Our behavioral AI platform was built from the ground up to model identity, behavior, and context across the enterprise. Combined with our AI-native development model, this foundation enables us to move quickly and extend intelligence into new security domains without the constraints of legacy architectures. Stephen will help lead this next phase, expanding our platform to solve a broader set of customer problems with speed and precision. The future of security is AI-native, adaptive, and behavior-driven. We’re excited to keep building it. Read more on our blog: abnormal.ai/blog/stephen-h…

Researchers at @AbnormalAI have uncovered a commercial-grade phishing platform called Starkiller that uses live reverse proxy attacks to steal credentials and bypass multi-factor authentication in real time. 🔗 Read more: informationsecuritybuzz.com/new-phishing-k… #ISBNews

🚨 New Phishing Framework: Starkiller Makes MFA Bypass Dangerously Simple Most phishing kits rely on static HTML clones of login pages. Starkiller doesn't. Using headless browsers and reverse proxies, it relays live authentication sessions—capturing credentials, MFA codes, and session tokens in real time while the target interacts with a legitimate site. In our blog post, we break down: • How a polished control panel makes enterprise-grade phishing accessible to low-skill attackers • Why MFA offers little protection when the full authentication flow is proxied in real time • How built-in URL masking tools make malicious links appear legitimate to recipients • What real-time session monitoring means for targets—and why they never know • How Starkiller is distributed like a SaaS product, with ongoing support and regular updates 💡 Why this matters: Because each phishing session is generated dynamically, there are no static artifacts for security tools to fingerprint or blocklist. Traditional defenses—static page analysis, domain blocklisting, reputation-based URL filtering—weren't built for this. Stopping these attacks requires behavioral detection at the inbox level, before the link ever reaches a user. ➡️ Read the full analysis: abnormal.ai/blog/starkille…

Legacy security can’t keep up with human-centric attacks. Behavioral AI can. We’re proud to be named to the 2026 @CRN Security 100 list. This recognition validates our mission: moving beyond static rules to protect the human behavior at the heart of every enterprise. Thank you to our partners and customers who trust us to secure their future. crn.com/news/security/… #CRNSecurity100

The 2026 threat landscape is here. Attackers are no longer just sending "bad" links; they’re exploiting identity and routine workflows. If your security doesn't understand human behavior, it doesn't understand the threat. ICYMI, read our 2026 Threat Forecast to see what’s ahead: 🔗 abnormal.ai/blog/threat-fo…

Experience Abnormal AI After Dark During RSAC 🏙️ ️️ We’re bringing our customers a lineup that can't be missed. Prepare for an epic night with exclusive performances by Kaskade and Busta Rhymes. This is an exclusive customer appreciation party, with the mission of bringing the cybersecurity community together for a memorable experience. Request an invitation to join us at The Warfield in San Francisco during RSAC week! abnormal.ai/rsac-party?utm…

Global teams deserve security training that actually speaks their language. Too often, translated awareness content feels generic—checking a box without changing behavior. And as phishing attacks evolve, manual translation simply can’t keep pace. AI Phishing Coach delivers adaptive simulations, just-in-time coaching, and AI-generated video training automatically in each employee’s preferred language—driving real engagement and measurable resilience worldwide. Read how AI is helping global organizations build a stronger, unified security culture at scale. 👇 abnormal.ai/blog/ai-multil…

We’re thrilled to see Abnormal recognized in @CSOonline's list of Top 10 Vendors for AI-Enabled Security—According to CISOs! This list reflects how CISOs are thinking about the future of security: not just who has the flashiest AI, but who delivers real innovation, strong reputation, business value, reasonable cost, seamless integration, and peer trust. Based on feedback from CISOs, this underscores how decision-makers in security are leaning toward vendors who can be trusted to solve real problems with smart, reliable technology. Check out the full article: csoonline.com/article/411578…