AI in Cybersecurity

AI is not failing because of bad ideas; it’s "failing" at enterprise scale because of two big gaps: 👉 Workforce Preparation 👉 Data Security for AI While I speak globally on both topics in depth, today I want to educate us on what it takes to secure data for AI—because 70–82% of AI projects pause or get cancelled at POC/MVP stage (source: #Gartner, #MIT). Why? One of the biggest reasons is a lack of readiness at the data layer. So let’s make it simple - there are 7 phases to securing data for AI—and each phase has direct business risk if ignored. 🔹 Phase 1: Data Sourcing Security - Validating the origin, ownership, and licensing rights of all ingested data. Why It Matters: You can’t build scalable AI with data you don’t own or can’t trace. 🔹 Phase 2: Data Infrastructure Security - Ensuring data warehouses, lakes, and pipelines that support your AI models are hardened and access-controlled. Why It Matters: Unsecured data environments are easy targets for bad actors making you exposed to data breaches, IP theft, and model poisoning. 🔹 Phase 3: Data In-Transit Security - Protecting data as it moves across internal or external systems, especially between cloud, APIs, and vendors. Why It Matters: Intercepted training data = compromised models. Think of it as shipping cash across town in an armored truck—or on a bicycle—your choice. 🔹 Phase 4: API Security for Foundational Models - Safeguarding the APIs you use to connect with LLMs and third-party GenAI platforms (OpenAI, Anthropic, etc.). Why It Matters: Unmonitored API calls can leak sensitive data into public models or expose internal IP. This isn’t just tech debt. It’s reputational and regulatory risk. 🔹 Phase 5: Foundational Model Protection - Defending your proprietary models and fine-tunes from external inference, theft, or malicious querying. Why It Matters: Prompt injection attacks are real. And your enterprise-trained model? It’s a business asset. You lock your office at night—do the same with your models. 🔹 Phase 6: Incident Response for AI Data Breaches - Having predefined protocols for breaches, hallucinations, or AI-generated harm—who’s notified, who investigates, how damage is mitigated. Why It Matters: AI-related incidents are happening. Legal needs response plans. Cyber needs escalation tiers. 🔹 Phase 7: CI/CD for Models (with Security Hooks) - Continuous integration and delivery pipelines for models, embedded with testing, governance, and version-control protocols. Why It Matter: Shipping models like software means risk comes faster—and so must detection. Governance must be baked into every deployment sprint. Want your AI strategy to succeed past MVP? Focus and lock down the data. #AI #DataSecurity #AILeadership #Cybersecurity #FutureOfWork #ResponsibleAI #SolRashidi #Data #Leadership

The National Institute of Standards and Technology (NIST) has released a draft of its “Cybersecurity Framework Profile for Artificial Intelligence” (open for public comment until Jan 30, 2026) to help organizations think about how to strategically adopt AI while addressing emerging cybersecurity risks that stem from AI’s rapid advance. Building on the #NIST Cybersecurity Framework 2.0, the Cyber AI Profile translates well-established risk management concepts into AI-specific cybersecurity considerations, offering a practical reference point as organizations integrate AI into critical systems and confront AI-enabled threats. The Cyber AI Profile centers on three focus areas: • Securing AI systems: identifying cybersecurity challenges when integrating AI into organizational ecosystems and infrastructure. • Conducting AI-enabled cyber defense: identifying opportunities to use AI to enhance cybersecurity, and understanding challenges when leveraging AI to support defensive operations. • Thwarting AI-enabled cyberattacks: building resilience to protect against new AI-enabled threats. The Profile complements existing NIST frameworks (CSF, AI RMF, RMF) by prioritizing AI-specific cybersecurity outcomes rather than creating a standalone regime.

Today, NIST released the initial preliminary draft of the Cybersecurity Framework Profile for Artificial Intelligence (Cyber AI Profile), a community profile built on NIST CSF 2.0 to help organizations manage cybersecurity risk in an AI-driven world. A key section of this draft is Section 2.1, which introduces three Focus Areas that explain how AI and cybersecurity intersect in practice: 1. Securing AI System Components (Secure) AI systems introduce new assets that must be secured; models, training data, prompts, agents, pipelines, and deployment environments. This focus area emphasizes treating AI components as first-class cybersecurity assets, integrating them into governance, risk assessments, protection controls, and monitoring processes. It reinforces that AI risk should not be siloed from enterprise cybersecurity risk management. 2. Conducting AI-Enabled Cyber Defense (Defend) AI is not just something to protect, it is also a powerful defensive capability. This area focuses on using AI to enhance detection, analytics, automation, and response across security operations. At the same time, it recognizes the risks of over-reliance on automation, model integrity concerns, and the need for human oversight when AI supports security decision-making. 3. Thwarting AI-Enabled Cyber Attacks (Thwart) Adversaries are increasingly using AI to scale phishing, evade detection, and automate attacks. This focus area addresses how organizations must anticipate and counter AI-enabled threats by building resilience, improving detection of AI-driven attack patterns, and preparing for a rapidly evolving threat landscape where AI is weaponized. Why This Matters Together, Secure, Defend, and Thwart provide a practical structure for aligning AI initiatives with existing cybersecurity programs. By mapping AI-specific considerations to CSF 2.0 outcomes (Govern, Identify, Protect, Detect, Respond, Recover), the Cyber AI Profile helps organizations integrate AI security into familiar risk management practices. This is a preliminary draft, and NIST is seeking public feedback through January 30, 2026. If your organization is building, deploying, or defending with AI, now is the time to review and contribute. 🔗 https://lnkd.in/e-ETZXH8

Leveraging this new OpenAI real time translator to phish via phone calls in the target’s preferred language in 3…2… So far, AI has been used for believable translations in phishing emails — E.g. my Icelandic customers are seeing a massive increase in phishing in their language in 2024. Before only 350,000 or so people comfortably spoke Icelandic correctly, now AI can do it for the attacker. We’re going to see this real time translation tool increasingly used to speak in the target’s preferred language during phone call based attacks. These tools are easily integrated into the technology we use to spoof caller ID, place calls, and voice clone. Now, in any language. Educate your team & family + friends. Make sure folks know: - AI can voice clone - AI can real time translate to speak in any language - Caller ID is easily spoofed with or without AI tools - AI tools will increase in believability Example AI voice clone/spoof example here: https://lnkd.in/gPMVDBYC Will this AI be used for good? Sure! Real time translations are quite useful for people, businesses, & travel. We still need to educate folks on how AI is currently use to phish people & how real time AI translations will increase scams across (previous) language barriers. *What can we do to protect folks from attackers using AI to trick?* - Educate first: make sure folks around you know it’s possible for attackers to use AI to voice clone, deepfake video and audio (in real time during calls) - Be politely paranoid: encourage your team and community to use 2 methods of communication to verify someone is who they say they are for sensitive actions like sending money, data, access, etc. For example, if you get a phone call from your nephew saying he needs bail money now, contact him a different way before sending money to confirm it’s an authentic request - Passphrase: consider using a passphrase with your loved ones to verify identity in emergencies (e.g. your sister calls you crying saying she needs $1,500 urgently ask her to say the passphrase you agreed upon together or contact with another communication method before sending money)

AI agents have security problems that most companies are not ready for. Not because agents are dangerous by default. Because agents need access. - To databases. - Customer records - Personnel files - Financial data. - Internal policies. - Workflows. - Permissions. That means the data layer is no longer just storage. It is becoming the new security boundary. This is the part many enterprise AI strategies miss. If your AI agent can query the data, summarize the data, reason over the data, and act on the data, then security cannot only sit in the application layer. It has to live where the data lives. That is why Oracle’s latest AI Database security push is interesting. The message is simple: Secure at the source. Secure at speed. Secure through resilience. Offer security, patching and upgrade tools at no cost or deeply discounted to get started fast. In other words: Protect the data directly. Patch faster than attackers move. Recover quickly when things go wrong. Get into an accelerated data protection cycle. That is the new AI security model. https://lnkd.in/gSynh72J The next AI questions are not just: “How many agents can we deploy?” It is: “Is our data secure enough to survive them? “Do we have a strategy to stop rogue agents and shadow agents at the source?” and  “Can our architecture scale to sustain hundreds, thousands or hundreds of thousands of AI agents?”

When AI Meets Security: The Blind Spot We Can't Afford Working in this field has revealed a troubling reality: our security practices aren't evolving as fast as our AI capabilities. Many organizations still treat AI security as an extension of traditional cybersecurity—it's not. AI security must protect dynamic, evolving systems that continuously learn and make decisions. This fundamental difference changes everything about our approach. What's particularly concerning is how vulnerable the model development pipeline remains. A single compromised credential can lead to subtle manipulations in training data that produce models which appear functional but contain hidden weaknesses or backdoors. The most effective security strategies I've seen share these characteristics: • They treat model architecture and training pipelines as critical infrastructure deserving specialized protection • They implement adversarial testing regimes that actively try to manipulate model outputs • They maintain comprehensive monitoring of both inputs and inference patterns to detect anomalies The uncomfortable reality is that securing AI systems requires expertise that bridges two traditionally separate domains. Few professionals truly understand both the intricacies of modern machine learning architectures and advanced cybersecurity principles. This security gap represents perhaps the greatest unaddressed risk in enterprise AI deployment today. Has anyone found effective ways to bridge this knowledge gap in their organizations? What training or collaborative approaches have worked?

The 6-Layer AI Security Stack Every Organization Will Eventually Need Most companies don't have an AI strategy. They have an AI chatbot. And those are not the same thing. The companies that will struggle with AI over the next few years won't be the ones with the weakest models. They'll be the ones with the weakest security. Because securing AI isn't about adding one tool. It's about building an entire security stack. Here's what a modern AI Security Stack looks like: 1. Identity & Access Layer Control who can access models, agents, APIs, and sensitive AI workflows. Without identity controls, anyone with access can become your biggest risk. 2. Data Protection Layer Protect sensitive information before it ever reaches an LLM. • Mask PII • Encrypt data • Tokenize sensitive fields If your prompts contain confidential data, your security starts before inference. 3. Prompt & Input Security Layer AI models trust their inputs. Attackers know that. Defend against: • Prompt injection • Jailbreak attempts • Data extraction attacks Every prompt should be treated as untrusted input. 4. Governance & Compliance Layer Security isn't only technical. It's also accountability. Track: • Risk classifications • Audit trails • AI decisions • Regulatory compliance AI without governance becomes impossible to trust at scale. 5. Output Validation Layer Never assume the model is right. Validate every critical response for: • Hallucinations • Policy violations • Compliance issues • Unsafe recommendations Trust... but verify. 6. Monitoring & Observability Layer Deployment isn't the finish line. It's where security actually begins. Continuously monitor: • Model drift • Unusual behavior • Performance changes • Security events • Response quality You can't defend what you can't observe. The biggest misconception about AI security? People think it's one product. In reality, it's multiple security layers working together. Just like cloud security evolved from firewalls to full security architectures... AI security is following the same path. The organizations building these layers today won't just deploy AI faster. They'll deploy it with confidence. AI is becoming part of every business. AI security needs to become part of every architecture. Which layer do you think organizations are overlooking the most right now? Follow Marcel Velica for practical insights on AI Security, Cybersecurity, and Enterprise AI. If you found this useful, repost it so more security professionals can join the conversation.

🇸🇬 [AI SECURITY] Singapore takes the lead in AI governance again! The Cyber Security Agency of Singapore (CSA) released AI security guidelines that EVERYONE developing or deploying AI should know: 1️⃣ Take a lifecycle approach "As with good cybersecurity practice, CSA recommends that system owners take a lifecycle approach to consider security risks. Hardening only the AI model is insufficient to ensure a holistic defence against AI related threats. All stakeholders involved across the lifecycle of an AI system should seek to better understand the security threats and their potential impact on the desired outcomes of the AI system, and what decisions or trade-offs will need to be made. The AI lifecycle represents the iterative process of designing an AI solution to meet a business or operational need. As such, system owners will likely revisit the planning and design, development, and deployment steps in the lifecycle many times in the delivery of an AI solution." 2️⃣ Start with risk assessment "Given the diversity of AI use cases, there is no one-size-fits-all solution to implementing security. As such, effective cybersecurity starts with conducting a risk assessment. This will enable organisations to identify potential risks, priorities, and subsequently, the appropriate risk management strategies. A fundamental difference between AI and traditional software is that while traditional software relies on static rules and explicit programming, AI uses machine learning and neural networks to autonomously learn and make decisions without the need for detailed instructions for each task. As such, organisations should consider conducting risk assessments more frequently than for conventional systems, even if they generally base their risk assessment approach on existing governance and policies. These assessments may also be supplemented by continuous monitoring and a strong feedback loop." 3️⃣ Guidelines for securing AI systems ⮕ "Planning and design → Raise awareness and competency on security risks  → Conduct security risk assessments ⮕ Development → Secure the supply chain  → Consider security benefits and trade-offs when selecting the appropriate model to use → Identify, track and protect AI-related assets → Secure the AI development environment ⮕ Deployment → Secure the deployment infrastructure and environment of AI systems → Establish incident management procedures → Release AI systems responsibly ⮕ Operations and Maintenance → Monitor AI system inputs → Monitor AI system outputs and behaviour → Adopt a secure-by-design approach to updates and continuous learning → Establish a vulnerability disclosure process ⮕ End of Life → Ensure proper data and model disposal" ➡️ Read the full report below (download the companion guide too). 🏛️ STAY UP TO DATE. AI governance is moving fast: join 36,700+ people who subscribe to my newsletter on AI policy, compliance & regulation (link below). #AI #AISecurity #AIGovernance #AIRisks

AI is changing the economics and speed of cyberattacks. What once took threat actors days or weeks can now happen in minutes: automated reconnaissance, AI-assisted exploit development, credential targeting, lateral movement, and highly personalized phishing at scale. This is why Palo Alto Networks believes so strongly in the concept of autonomous resilience. The traditional model of security operations: fragmented tools, manual escalation paths, and human-speed response cycles - was not designed for machine-speed threats. Autonomous resilience means building security architectures that can continuously reduce exposure, validate trust, and contain threats in real time. What does that look like in practice? 🔸 Minimize attack surface Continuously identify and remediate exposed assets, misconfigurations, vulnerable APIs, and unmanaged cloud resources before attackers can weaponize them. For example, AI-driven exposure management can detect an internet-facing development environment created outside policy and trigger automated remediation immediately. 🔸 Secure every identity Trust must extend beyond employees to machine identities, workloads, APIs, and AI agents. This means enforcing least privilege, adaptive access controls, and continuous identity validation to stop credential misuse and token theft before attackers gain persistence. 🔸 Defend the software supply chain AI-assisted attacks increasingly target CI/CD pipelines, open-source dependencies, and code repositories. Organizations need runtime protections, code integrity validation, and automated policy enforcement to prevent manipulated code from reaching production environments. 🔸 Constrain blast radius Zero Trust architectures become even more critical in an AI-driven threat landscape. Microsegmentation, continuous inspection, and behavioral analytics help prevent attackers from moving laterally across environments once initial access is achieved. 🔸 Detect and respond in real time Security teams cannot rely on analysts manually correlating thousands of alerts. AI-driven SOC operations can automatically prioritize incidents, enrich telemetry, isolate compromised assets, and initiate containment workflows within minutes — dramatically reducing operational fatigue and response time. The outcome is not “fully autonomous security.” The outcome is resilient organizations that can adapt, contain, and recover faster in an increasingly automated threat environment. Cybersecurity is evolving from reactive defense into continuous operational resilience. The organizations preparing for that shift now will be far better positioned for what comes next.

We all know AI will continue to be the defining conversation for 2026, but what I’m hearing most often from leaders is: “How do we leverage AI without introducing untenable risk?” This year, we will see three defining shifts, all underpinned by the top priority for the CEO and the critical operational mandate for the CIO: security. AI is transforming the threat landscape faster than most organizations can adapt, and a reactive approach is a business risk. An AI-powered defense shield is the foundation for safe reinvention. It’s about real-time visibility, actionable insights, and closing the loop from discovery to remediation across IT, OT, and cloud silos. This strategic and operational imperative shapes our three key shifts: 📌 Proliferation of (Secure) AI Agents: Beyond chatbots to specialized agents embedded in every function - HR, IT, customer service - running autonomous workflows. They become proactive partners, but every connected asset they touch expands the attack surface. The CIO's mandate: ensure this happens securely, at scale. 📌 Deepening Industry Impact with Real-Time Protection: True transformation happens in mission-critical workflows. In healthcare, with thousands of connected devices managing patient data. In manufacturing, on smart factory floors. The CEO needs confidence that business reinvention can happen in their industry; the CIO needs a unified platform to see, decide, and act across it all. 📌 Expanding a Unified Security Posture: Our “ANY” strategy - connecting to any model, any data, any service - demands a unified view of risk. Observability, asset management, incident response… Risk doesn’t stay in silos; to manage it requires architecture that breaks down walls between IT, security, and operations. This is the year intelligent, secure automation becomes inseparable from business strategy. The organizations that thrive will be those that align the CEO's security-first vision with the CIO's execution, proactively seeing every asset, prioritizing every risk, and acting before an incident occurs. Here’s to a transformative - and secure - 2026. #AI #CyberSecurity #DigitalTransformation