Blog

Reflections from RSAC 2026, San FranciscoThe message at this year's RSA Conference was impossible to ignore. It came from keynote stages, panel discussions, and casual conversations on the show floor, and it was always the same: you cannot fight machine-speed attacks with human-speed defense.That's not a philosophical position anymore. It's a measurable, documented operational reality.Consider what the numbers tell us. The average breach hand-off time - the window between initial compromise and lateral movement - collapsed from 8 hours in 2022 to just 22 seconds in 2025. Ransomware breakout now happens in under a minute. And perhaps most unsettling of all, 80% of successful intrusions today involve zero malware. No executable. No payload to detect. Just stolen credentials moving quietly through systems that were never designed to question them.Attackers didn't wait for the security industry to catch up. They adopted AI first, and they're using it well - probing networks at superhuman scale, generating endless phishing variants tailored to individual targets, and exploiting identities in ways that leave almost no traditional trace.The Industry Responds: Enter the Autonomous SOCThe security industry's answer to all of this is the Autonomous Security Operations Center - AI agents capable of triaging alerts, investigating incidents, and initiating remediation before a human analyst has even opened a ticket.At RSAC 2026, this wasn't theoretical. Google, CrowdStrike, and Datadog all unveiled their versions of the Autonomous SOC. The architecture is compelling: agents that operate continuously, correlate signals at scale, and compress what used to take hours of human analysis into seconds of automated response.In a threat environment defined by 22-second breakout windows, that compression isn't a nice-to-have. It's the only mathematically viable defense.Three Hard Truths Behind the HypeBut for all the announcements and demos, RSAC 2026 also surfaced three uncomfortable realities that no vendor keynote fully addressed.1. Non-human identities are the new perimeter.In most modern enterprises, bots, AI agents, and service accounts now outnumber human accounts. These identities authenticate, access data, execute tasks, and communicate with external systems - and the vast majority of them are poorly governed, inconsistently monitored, and completely invisible to traditional identity security programs. When attackers want a way in, they increasingly don't bother with humans at all.2. Shadow agents are a live threat.Employees across organizations are deploying unapproved AI tools - for writing, for analysis, for productivity - without the knowledge of their security teams. Every one of these tools represents a potential data exfiltration path that sits entirely outside existing monitoring frameworks. The problem isn't malicious intent; it's the structural invisibility of it. Shadow AI is the shadow IT problem of the 2020s, and it's moving faster.3. Deployment lag is the real vulnerability.Here is the gap that matters most. Nine in ten organizations claim to use AI in their security stack. Yet three out of four apply it to less than 10% of their actual portfolio. The tools exist. The technology is mature. The problem is organizational will and deployment discipline. Attackers don't have governance committees. Defenders do. That asymmetry is being exploited in real time.The Train Has Left the StationOne analyst at the conference put it with unusual directness: "You need AI to secure AI. If you're a security team not on board with AI, you will fall behind very quickly. There's no stopping the train."That's a blunt framing, but it's the accurate one. The AI arms race in cybersecurity is not a metaphor or a marketing narrative. It is the operational reality of 2026. The organizations that close the deployment gap - that govern their non-human identities, shine light on their shadow agents, and actually scale their AI security tools beyond the pilot stage - are the ones that will remain defensible.The window to get ahead of this is still open. But it has never been smaller.What are you seeing on the AI security front in your organization? I'd like to hear how teams are navigating the gap between having the tools and actually deploying them at scale.

AI vs AI: The New Arms Race in Cybersecurity

a computer generated image of the letter a

AI agent security is the practice of protecting both AI agents and the systems they interact with from misuse, manipulation, and exploitation. As organizations increasingly deploy agentic AI to automate decisions, execute tasks, and interact with sensitive systems, the attack surface grows significantly. Securing AI agents is no longer just about model safety; it is about ensuring these agents operate as intended, do not expose confidential data, and cannot be weaponized for harmful purposes.Recent incidents have shown why this is such a pressing concern. In one reported case, a Meta AI agent malfunction allegedly exposed sensitive company and user data to employees who were not authorized to access it. In the same incident, an employee reportedly relied on guidance from the AI agent that should not have been followed. Without fully analyzing the consequences, the employee acted on the advice, leading to a security incident. Meta reportedly classified the event as “Sev1,” indicating a highly serious internal security issue.This was not said to be an isolated example. Last summer, Meta Superintellignece’s safety and alignment director reportedly shared that the OpenClaw agent deleted her entire inbox even though she had explicitly instructed it to confirm before taking any action. Incidents like these highlight a key reality of agentic AI: the danger is not always a malicious external attacker. Sometimes the threat comes from the agent making incorrect decisions, overstepping its permissions, or being trusted too much by the humans using it.The threat landscape for AI agents is multifaceted. Unlike traditional applications, AI agents can process natural language, access tools, call APIs, store memory, and take actions across connected systems. This creates a wide range of vulnerabilities that attackers can exploit and that organizations must proactively defend against. Key AI Agent Security VulnerabilitiesSome of the most important vulnerabilities in agentic AI systems include:Prompt injection Attackers manipulate the agent through crafted inputs that override system instructions or cause unintended behavior.Tool and API manipulation If an agent has access to tools or APIs, attackers may exploit weak validation or poor controls to force unauthorized actions.Data poisoning Malicious or corrupted data introduced into training or retrieval systems can distort agent behavior and decision-making.Memory poisoning Attackers can influence the agent’s memory or stored context, causing harmful or misleading actions in future interactions.Privilege compromise Agents with excessive permissions can become dangerous if they are manipulated or malfunction.Authentication and access control spoofing Weak identity checks may allow attackers to impersonate authorized users or services.Remote code execution (RCE) attacks In poorly secured environments, an agent’s interaction with code execution tools can create pathways for system compromise.Cascading failures and resource overload One compromised or malfunctioning agent can trigger larger operational failures across interconnected systems. Why AI Agent Security Is DifferentTraditional cybersecurity focuses on protecting systems from unauthorized access, malware, and network-based attacks. AI agent security goes a step further. It must also address decision integrity, instruction reliability, contextual understanding, and action safety. An agent does not need to be “hacked” in the conventional sense to create damage. A misleading prompt, a poisoned memory, or poorly bounded permissions may be enough to trigger a serious incident.This makes AI agents both powerful and risky. They can accelerate productivity, but they can also amplify mistakes at machine speed. The more autonomous the system becomes, the more important strong guardrails, validation layers, and access controls become. Best Practices for Securing AI AgentsDespite the wide and varied threat landscape, agentic AI systems can be secured through effective countermeasures and strong AI guardrails. Organizations that adopt a proactive security posture and follow current best practices for vulnerability management will be better positioned to reduce risk and stay ahead of increasingly sophisticated cyber threats.Key best practices include:Zero trust architecture Never assume trust by default. Every request, identity, and system interaction should be continuously verified.Principle of least privilege Give agents only the minimum level of access they need to perform their tasks. This reduces blast radius if something goes wrong.Context-aware authentication Authentication should consider not just identity, but also device, location, behavior, and request context.Data encryption Sensitive data should be encrypted both at rest and in transit to reduce the risk of exposure.Microsegmentation Isolate systems and workloads so that even if one component is compromised, the attacker cannot move freely across the environment. Prompt hardening Design system prompts carefully to make them more resistant to manipulation and instruction override.Prompt validation Validate and sanitize inputs before they influence agent behavior, especially when external or untrusted input is involved. The Path ForwardAs enterprises continue adopting agentic AI, security must be built into the design from the beginning rather than added later as an afterthought. AI agents should not be treated as simple productivity tools. They are autonomous or semi-autonomous actors that interact with data, systems, and people in ways that can create both value and risk.The future of AI in the enterprise depends not only on what agents can do, but on how safely they can do it. Organizations that invest in robust AI agent security today will be better prepared to harness the benefits of automation without exposing themselves to preventable breaches, data leaks, and operational failures.In the end, AI agent security is about trust. If agents are going to make decisions, access systems, and take actions on behalf of humans, they must be governed by strong controls, clear boundaries, and continuous oversight. Without that, even a helpful agent can quickly become a serious security liability.

AI Agent Security: Why It Matters More Than Ever

E-mail spam phishing scammer concept. Red warning symbol on envelope in fishing hook icon as bait on laptop screen, criminal hacker cyberattack sending scam malware spreading virus sms text message.

La computación en la nube continúa transformando la manera en que operan las organizaciones, pero con ese crecimiento también se amplía la superficie de ataque, especialmente en la capa de identidad. Una reciente campaña maliciosa que impactó a múltiples entornos en la nube de Azure demuestra cómo los atacantes están cambiando su enfoque de la explotación de la infraestructura a la explotación de cuentas en la nube (Cloud Account Takeover – ATO) mediante phishing y abuso de identidad.La campaña comenzó con correos electrónicos de phishing disfrazados de documentos compartidos en la nube. Las víctimas que hicieron clic en los enlaces fueron redirigidas a páginas de captura de credenciales que imitaban los portales de inicio de sesión legítimos. Una vez que los atacantes obtuvieron credenciales válidas, ingresaron rápidamente a las cuentas comprometidas, añadieron sus propios métodos de autenticación multifactor (MFA) y establecieron una persistencia a largo plazo. A partir de ahí, las cuentas comprometidas se utilizaron para enviar nuevos mensajes de phishing, tanto internos como externos, acceder a datos sensibles en la nube y, en algunos casos, intentar fraudes financieros. Los atacantes también crearon reglas de correo para ocultar alertas de seguridad y borrar rastros del compromiso.Esta campaña demuestra una realidad crítica en los entornos modernos en la nube: la identidad se ha convertido en la principal superficie de ataque. Las técnicas observadas, como el robo de credenciales, la manipulación de MFA, el abuso de buzones y el acceso mediante proxies, no son exclusivas de un solo proveedor de nube. Cualquier organización que dependa de la gestión de identidades y accesos en la nube es potencialmente vulnerable a ataques similares.Para la comunidad de la Cloud Security Alliance (CSA), este incidente refuerza varios temas clave sobre el riesgo en la nube. En primer lugar, el phishing sigue siendo uno de los puntos de entrada más eficaces para los atacantes. En segundo lugar, el uso de MFA por sí solo no es suficiente si las organizaciones no supervisan activamente los cambios sospechosos en los métodos de autenticación. Por último, sin registros continuos ni monitoreo del comportamiento, los atacantes pueden operar en entornos en la nube durante largos periodos sin ser detectados.Recomendaciones Clave de Defensa Para reducir la exposición a ataques de identidad en la nube, las organizaciones deben priorizar las siguientes medidas defensivas:<ul><li>Implementar MFA para todos los usuarios y supervisar activamente inscripciones o cambios no autorizados en los métodos de autenticación</li><li>Aplicar el principio de mínimo privilegio para limitar el impacto de una cuenta comprometida</li><li>Monitorear comportamientos anómalos de inicio de sesión, nuevas reglas de correo y actividades inusuales de agentes de usuario</li><li>Fortalecer los controles de seguridad en el correo electrónico y las plataformas de colaboración en la nube</li><li>Realizar auditorías periódicas de seguridad en la nube y revisiones de acceso</li><li>Proporcionar capacitación continua de concienciación en seguridad enfocada en técnicas modernas de phishing, especialmente aquellas que utilizan documentos compartidos y herramientas de colaboración</li></ul>Esta campaña sirve como un recordatorio contundente de que la seguridad en la nube no se limita a proteger máquinas virtuales y redes; también implica proteger identidades, rutas de acceso y el comportamiento de los usuarios. A medida que la adopción de la nube continúa acelerándose, las organizaciones deben dar mayor prioridad a la gobernanza de identidades, al monitoreo continuo y a la educación de los usuarios para reducir el riesgo de secuestro de cuentas en la nube.

Una Llamada de Atención para la Seguridad en la Nube: Lecciones de una Reciente Campaña de Secuestro de Cuentas en Azure

Cloud computing continues to transform how organizations operate, but with that growth comes an expanding attack surface, especially at the identity layer. A recent malicious campaign targeting multiple Azure cloud environments highlights how attackers are shifting their focus from infrastructure exploitation to cloud account takeover (ATO) through phishing and identity abuse.The campaign began with phishing emails disguised as shared cloud documents. Victims who clicked the links were redirected to credential-harvesting pages that closely mimicked legitimate login portals. Once attackers obtained valid credentials, they quickly logged into victim accounts, added their own multi-factor authentication (MFA) methods, and established long-term persistence. From there, compromised accounts were used to send additional phishing messages internally and externally, access sensitive cloud data and, in some cases, attempt financial fraud. Attackers also created mailbox rules to hide security alerts and erase signs of compromise.This campaign demonstrates a critical reality of modern cloud environments: identity is now the primary attack surface. The techniques observed, credential theft, MFA manipulation, mailbox abuse, and proxy-based access, are not exclusive to one cloud provider. Any organization that relies on cloud-based identity and access management is potentially vulnerable to similar attacks.For the Cloud Security Alliance (CSA) community, this incident reinforces several important cloud risk themes. First, phishing remains one of the most effective entry points for attackers. Second, MFA alone is not enough if organizations are not actively monitoring for suspicious changes to authentication methods. Finally, without continuous logging and behavioral monitoring, attackers can operate inside cloud environments for extended periods without detection.Key Defensive Takeaways To reduce exposure to identity-driven cloud attacks, organizations should prioritize the following defensive measures:<ul><li>Enforce MFA across all users while actively monitoring for unauthorized MFA enrollment or changes</li><li>Apply least-privilege access to limit the blast radius of compromised accounts</li><li>Monitor for anomalous sign-in behavior, new mailbox rules, and unfamiliar user-agent activity</li><li>Strengthen email and cloud collaboration security controls to reduce phishing exposure</li><li>Perform regular cloud security audits and access reviews</li><li>Provide continuous security awareness training focused on modern phishing techniques involving shared documents and collaboration tools</li></ul>This campaign serves as a strong reminder that cloud security is not just about protecting virtual machines and networks — it is about securing identities, access paths, and user behavior. As cloud adoption continues to accelerate, organizations must place greater emphasis on identity governance, continuous monitoring, and user education to reduce the risk of cloud account takeover.

A Wake-Up Call for Cloud Security: Lessons from a Recent Azure Account Takeover Campaign

Human Digital Avatar as a Symbol of AI presses a virtual button on screen with Shield icon. Cyber Security and Data Protection Concept

The role of the Chief Information Security Officer has never been easy, but in 2025, it has fundamentally transformed. CISOs are being asked to secure a rapidly expanding digital ecosystem while managing tighter budgets, rising regulatory expectations, and an explosion of AI adoption across the enterprise.HackerOne’s latest research captures this shift with striking clarity:84% of CISOs are now responsible for AI security 82% are responsible for data privacy This expanded mandate creates more pressure than ever, especially as AI evolves faster than traditional security programs can keep pace.According to the Stanford 2025 AI Index, 78% of organizations used AI in 2024, a dramatic jump from 55% the year before. As AI tools are deployed independently by various business units, new and unpredictable attack surfaces are emerging almost overnight. Even the most talented internal security teams cannot be experts in every emerging AI risk, prompting CISOs to rethink their approach.&nbsp; Why CISOs Don’t Need Bigger Teams—They Need the Right Partner In conversations with CISOs across industries, a recurring theme emerges: today’s challenges cannot be solved with headcount alone. What leaders need is specialized, external expertise to uncover blind spots and validate emerging attack surfaces. This is where offensive security and crowdsourced expertise become essential.HackerOne’s research shows:88% of CISOs say crowdsourced security is effective in identifying data privacy vulnerabilities 81% say it helps detect AI-related threats This is no longer a niche tactic. Crowdsourcing is becoming a core component of modern security strategies, accelerating the shift toward Continuous Threat Exposure Management (CTEM), a move from periodic testing to continuous, risk-based validation.&nbsp; The “15% Advantage”: What High-Performing CISOs Do Differently Among the CISOs surveyed, a standout group, the top 15%, has fully embraced a comprehensive crowdsourced security strategy. These leaders are: 2x more likely to report strong effectiveness from their programs Leveraging multiple offensive security layers, including:Bug bounty programs Vulnerability disclosure programs (VDPs) Red teaming Pentesting AI-assisted security workflows By combining AI with the diverse expertise of global security researchers, this group uncovers vulnerabilities across the entire software development lifecycle including emerging AI risks and new privacy exposures. Bug bounty programs, in particular, offer a pay-for-results model, giving CISOs a cost-effective way to secure expanding attack surfaces without ballooning internal budgets.&nbsp; Closing the Cyber Talent Gap At Scale Another major obstacle facing today’s CISOs is the ongoing security talent shortage: 39% cite lack of skilled personnel as a key challenge The global workforce is short four million cybersecurity professionals, according to the World Economic Forum Crowdsourced security delivers a scalable solution to this talent gap. Instead of relying solely on limited internal expertise, organizations gain on-demand access to thousands of vetted, highly specialized researchers. This diversity across geographies, backgrounds, and skill sets provides coverage that is nearly impossible to replicate in-house. It allows organizations to match the speed and complexity of modern threats without being constrained by traditional hiring bottlenecks.&nbsp; Future-Proofing Security Through Crowdsourced Expertise HackerOne’s research found a striking insight: 100% of CISOs who fully embrace crowdsourced security see it as critical to their overall strategy. For boards and executive leaders, the message is clear. Crowdsourced security is no longer optional; it is becoming a strategic imperative. It transforms security from a reactive function into a proactive, resilient, and intelligence-driven advantage.As AI adoption accelerates and data privacy expectations intensify, organizations that leverage the global security community will be better positioned to detect emerging threats, continuously validate risks, and build a security posture designed for the future.

Why CISOs Are Turning to Crowdsourced Security to Protect AI and Data Privacy in 2025

El rol del Chief Information Security Officer (CISO) nunca ha sido sencillo, pero en 2025 ha experimentado una profunda transformación. Hoy, los CISOs deben asegurar un ecosistema digital en expansión, gestionar presupuestos más ajustados, cumplir con regulaciones cada vez más estrictas y proteger un uso de la IA que crece a un ritmo vertiginoso.La investigación más reciente de HackerOne evidencia este cambio de forma contundente:84% de los CISOs ahora son responsables de la seguridad de IA 82% son responsables de la privacidad de datos Este mandato ampliado intensifica la presión, especialmente dado que la IA evoluciona más rápido de lo que los programas de seguridad tradicionales pueden seguir.Según el AI Index 2025 de Stanford, el 78% de las organizaciones usaron IA en 2024, un salto drástico desde el 55% del año anterior. Con equipos de negocio que adoptan herramientas de IA sin una supervisión centralizada, surgen nuevas superficies de ataque prácticamente de la noche a la mañana. Incluso los equipos internos más talentosos no pueden abarcar todos los riesgos emergentes, lo que impulsa a los CISO a replantear su estrategia.&nbsp; Por qué los CISOs no necesitan equipos más grandes sino al socio adecuado En conversaciones con CISOs de diversas industrias, se observa un patrón claro: los desafíos actuales no se resuelven únicamente contratando más personal. Los líderes necesitan experiencia externa especializada para identificar puntos ciegos y validar nuevas superficies de ataque. Aquí es donde la seguridad ofensiva y la experiencia crowdsourced se vuelven esenciales.La investigación de HackerOne revela:88% de los CISOs encuentran efectiva la seguridad crowdsourced para identificar vulnerabilidades de privacidad de datos 81% aseguran que ayuda a detectar amenazas relacionadas con IA Esto ya no es una táctica aislada. El crowdsourcing se está convirtiendo en un componente central de las estrategias de seguridad modernas, impulsando la transición hacia Continuous Threat Exposure Management (CTEM), un modelo de validación continua y basado en riesgos, en lugar de pruebas periódicas.&nbsp; La “ventaja del 15%”: qué hacen diferente los CISOs de mayor rendimiento Dentro del estudio, un grupo destaca: el 15% superior de CISOs que ha adoptado por completo una estrategia crowdsourced integral. Estos líderes: son 2 veces más propensos a reportar alta efectividad en sus programas Aprovechan múltiples capas de seguridad ofensiva, incluyendo:Programas de bug bounty Programas de divulgación de vulnerabilidades (VDP) Red teaming Pentesting Flujos de trabajo de seguridad asistidos por IA Al combinar IA con la diversidad del talento global de investigadores, este grupo detecta vulnerabilidades en todo el ciclo de desarrollo de software, incluyendo riesgos emergentes de IA y exposiciones de privacidad. Los bug bounties, además, ofrecen un modelo de pago por resultados, lo que los convierte en una estrategia rentable para asegurar superficies de ataque cada vez más complejas sin inflar el presupuesto interno.&nbsp; Cerrar la brecha de talento—sin depender únicamente de contrataciones Otro obstáculo crítico para los CISOs es la escasez continua de talento en ciberseguridad:39% citan la falta de personal calificado como un desafío principal Faltan cuatro millones de profesionales a nivel global para cubrir los roles actuales, según el Foro Económico Mundial La seguridad crowdsourced proporciona una solución escalable para cerrar esta brecha. En lugar de depender exclusivamente de la experiencia interna limitada, las organizaciones acceden bajo demanda a miles de investigadores especializados y verificados. Esta diversidad de habilidades, enfoques y experiencias brinda un nivel de cobertura casi imposible de replicar internamente. Permite que las organizaciones igualen la velocidad y la complejidad de las amenazas modernas sin verse restringidas por los retos de contratación.&nbsp; Cómo blindar la seguridad hacia el futuro Un hallazgo contundente del estudio de HackerOne: 100% de los CISOs que adoptan plenamente la seguridad crowdsourced la consideran crítica para su estrategia general. Para los directorios y los líderes ejecutivos, el mensaje es claro. La seguridad crowdsourced ya no es opcional; es un imperativo estratégico. Transforma la seguridad de reactiva a una proactiva, resiliente y basada en inteligencia. A medida que la adopción de la IA se acelera y las expectativas de privacidad se intensifican, las organizaciones que aprovechen la comunidad global de investigadores estarán mejor posicionadas para detectar amenazas emergentes, validar riesgos de forma continua y construir una postura de seguridad preparada para el futuro.

Blog

AI vs AI: The New Arms Race in Cybersecurity

AI Agent Security: Why It Matters More Than Ever

Una Llamada de Atención para la Seguridad en la Nube: Lec...

A Wake-Up Call for Cloud Security: Lessons from a Recent ...

Why CISOs Are Turning to Crowdsourced Security to Protect...

Por qué los CISOs están adoptando la seguridad crowdsourc...