AI dominates cybersecurity experts’ 2026 predictions as it touches on authentication, privacy and attack acceleration.
GPU optimization and AI-driven attacks will push companies toward regional cloud providers
“GPU optimization becomes a headline topic in 2026. Today, most companies only use about 60% of the GPU power for which they are paying. Next-gen optimization software is going to flip that on its head, giving organizations the ability to squeeze full value out of their infrastructure. That matters not just for cost control, but for AI reliability. When your model performance becomes a competitive advantage, you can’t afford wasted compute, unpredictable throttling, or hardware carved into fractional units you can’t see. This is where optimized IaaS and regional GPU clouds start to shine.
“At the same time, attackers are getting smarter, and they’re starting to use AI too. The largest, most complex cloud environments become the biggest targets – when bad actors can spin up their own LLMs. Hyperscalers have hundreds of thousands of tenants, which means hundreds of thousands of potential attack surfaces (and pockets to pick). Regional providers have tighter vetting, cleaner environments, and fewer noisy neighbors.
“In 2026, security-conscious organizations will realize that the safest place to run AI and high-value workloads often isn’t the biggest cloud, it’s the one that actually keeps out the wrong people.”
– Richard Copeland, CEO of Leaseweb USA
AI-driven vulnerabilities and accelerated cyberattacks
“AI tools, such as ChatGPT, often store chat histories in the browser’s local storage, making sensitive conversations vulnerable to info-stealers. Despite warnings, many users continue to share sensitive topics with AI. While attackers will increasingly target such information, AI companies also use user data to train their models.
“2026 will also see a dramatic escalation in AI-powered offence and defence. AI has altered the accessibility and sophistication of cybercrime, lowering barriers for less technical actors while amplifying the capabilities of experienced criminals,’ said CTO Marijus Briedis.
“Cybercriminals are already experimenting with autonomous AI systems that can probe networks, identify weaknesses, and exploit vulnerabilities with minimal human oversight. These systems can learn, iterate, and adapt, making attacks faster and harder to predict, supporting phishing campaigns or social engineering. Advanced AI models like ‘Evil GPT’ are easily and cheaply available on the dark web, often for around $10.”
– NordVPN
Erosion of trust
“Trust is expected to become one of the biggest security challenges in 2026. As more services become fully cloud-based, authentication processes will be increasingly targeted. This includes deepfakes, voice cloning, realistic synthetic personas, automated phishing chats, and hyper-personalized attacks that blur the line between authentic and artificial.
“Criminals will create entirely fake synthetic identities, combining real user data with fabricated information, to access cloud accounts, open bank accounts, apply for credit, and commit crimes for years before detection. AI-enabled scams and fraud will increase productivity for criminals and make fraudulent websites and services increasingly difficult to detect. Ultimately, trust in digital devices and services may erode completely.”
– NordVPN
Staying ahead of AI-driven fraud: Advanced liveness and layered defence strategies
“AI has made it alarmingly easy to create fraudulent digital identities. The once-reliable barrier of biometric authentication is being compromised through camera injection attacks, where AI-generated or manipulated images are inserted into live video streams to deceive even the most sophisticated security systems.
“We are now entering an era defined not just by AI-driven fraud, but by a perpetual arms race between adversarial AI and defensive AI. Instead of relying on static or layered defences alone, businesses must adopt autonomous, connected identity defence systems that continuously learn, adapt, and collaborate across ecosystems. These intelligent systems both fight fraud and predict and prevent it, sharing insights across networks and adapting in real time to emerging attack vectors.
“Multimodal liveness detection, including combining visual, auditory, and motion-based signals, will remain critical, but its real power lies in how it integrates into a broader identity intelligence framework. This includes leveraging cross-customer fraud intelligence, behavioral biometrics, and transaction risk analytics to identify patterns of fraud before they manifest.
“In 2026 and beyond, the most resilient businesses will embrace a continuous, connected, and predictive fraud detection model. One that transforms AI from a reactive tool into a collaborative shield, capable of learning from every signal and staying one step ahead of even the most sophisticated fraudsters.”
– Ashwin Sugavanam, VP of AI & identity analytics, Jumio
AI and privacy at the core of future fraud prevention
“AI agents are making fraud more accessible and personalized than ever before. AI agents are lowering the barrier to executing complex fraud schemes, making it easier for fraudsters to automate attacks.
“To combat such AI-driven threats, businesses need a multi-layered approach. To address the worsening deepfake crisis, they must implement real-time solutions like multimodal liveness detection combined with rich contextual intelligence.
“Additionally, to protect user data during verification and build trust, companies should leverage privacy-preserving technologies like zero-knowledge proofs to combat identity fraud. By securely verifying identities without revealing sensitive data, companies can maintain user trust while effectively combating fraud.
“As we look ahead to 2026, fraud prevention in the digital space will be all about balancing security with user experience. Intelligent friction will become a critical strategy for businesses to address this challenge. By leveraging AI and machine learning, companies can tailor verification steps to a user’s risk profile and behaviour. Low-risk users will experience a smoother, less intrusive process, while high-risk users will be flagged for additional scrutiny.
“This approach will help organizations meet compliance and security requirements without alienating legitimate users. Much like how safe drivers benefit from lower insurance rates, those with a history of low-risk behavior will face fewer barriers, while suspicious activities will trigger more rigorous verification steps. This will help enterprises maintain security without compromising trust.”
– Alix Melchy, VP of AI at Jumio
Focus should shift from locking devices
“For years, cybersecurity has focused on locking down devices. We’ve wrapped them in management software, antivirus, and access controls, all in an attempt to contain data that should never have been there in the first place.
“Every year, enterprises pay more and more for corporate devices under the assumption they’re keeping corporate data separate from users’ personal devices. However, with apps like Outlook, Excel, and Google Sheets all accessible on mobile devices, a breach of a personal device is a breach of enterprise data.
“Moving into next year, AI-generated exploits will continue to be created and deployed in minutes. Our mobile device-driven society has extended the attack surface beyond the control of IT and cybersecurity staffs. We must concentrate on reducing the attack surface and protect our proprietary, sensitive and personal data with the same level of care.”
– Matt Stern, CSO of Hypori
AI outages become new “ransomware moment”
“In 2026, the biggest wake-up call for enterprises will be unexpected AI outages. As more organizations rely on AI systems for customer service, fraud detection, claims processing, supply chain routing, and decision automation, even a few minutes of downtime will create real-world business disruption. We’re moving into an era where AI is fully embedded into workflows, which means the databases, pipelines, and connections behind those AI systems must be architected for continuous availability. The companies that treat AI like a traditional app are going to run into the same wall we saw with ransomware years ago: you don’t realize how fragile the architecture is until it breaks.
What I’m seeing going into 2026 is a shift from ‘How do we deploy AI?’ to ‘How do we keep AI running, resilient, and trustworthy every second?’ The winners will be the companies that build durable foundations – resilient failover, airtight DR strategies, and secure, persistent connections between every environment where the data and compute live. AI will only be as reliable as the infrastructure supporting it. Businesses have to treat availability and security as non-negotiable if they want AI to successfully transform outcomes.”
– Don Boxley, co-founder and CEO of DH2i
KYA protocols become mandatory for high-value agentic commerce
“By the end of 2026, the financial and commercial liability shift caused by anonymous, autonomous agent fraud will force the industry to establish cryptographically mandated Know Your Agent (KYA) protocols for high-value transactions.
“Specifically, most top-tier global payment processors and wealth management platforms will require an evidence-grade, identity-bound payment token that is supported by an immutable consent audit trail. That will become the necessary step for any agent-delegated transaction exceeding a set risk threshold. This mandate will enforce least-privilege access via a token vault and demand verifiable human consent for agent actions, effectively transferring dispute liability to any party unable to produce cryptographic proof of authorization.”
– Mary Ann Miller, VP, evangelist, and fraud executive advisor at Prove
AI hardware shortages to define markets in 2026
“The pace of AI infrastructure expansion is colliding with physical limits in global supply chains, creating shortages of critical components and forcing markets to confront a new reality. This imbalance is already pushing up prices, concentrating market gains, and raising the risk of volatility.
“In 2026, this dynamic is likely to move from background noise to centre stage for investors. Supplies of several of these components are tight, with prices for advanced memory and specialized silicon rising sharply as manufacturers prioritize large AI customers.
“This is where the AI story turns hard-edged. It becomes about bottlenecks, allocation, and who controls supply. When scarcity appears, pricing power moves instantly. The companies sitting on essential components can protect margins and dictate terms. Everyone else absorbs the pressure.
“This concentration matters. Markets look stronger on the surface, but underneath, they’re becoming increasingly fragile. In 2026, this divergence is likely to be more pronounced as AI-related demand scales further and supply constraints persist.
“Investors may face an environment where inflation looks under control in aggregate, yet pricing pressure remains intense in precisely the sectors driving capital expenditure and market leadership. AI is not lifting all boats; it is selecting winners and losers.”
“This raises the stakes. When markets are built around constrained supply, miscalculations get priced in faster. The days of vague AI exposure are over. Precision matters.
“AI demand is real and durable, but scarcity, among other factors such as performance over hype, will now be driving the market narrative. Investors ignoring that shift do so at their own risk.”
– deVere Group CEO Nigel Green
AI becomes truly agentic
“AI is no longer just a tool for optimization. In 2026, agentic AI starts replacing full workflows, and that shift will separate companies that understand how to use AI from those that fight it. The real impact isn’t that AI replaces jobs, but that it replaces the tasks people shouldn’t be doing in the first place – the repetitive, time-sucking operations that drain teams. Organizations that lean into agentic AI will run faster, make decisions earlier, and redirect people into work that actually moves the business.
“As AI becomes more embedded in day-to-day operations, more companies will realize that complexity and cost are pushing them away from the hyperscalers. They’re seeing outages, noisy neighbor issues, unpredictable billing, and environments so complex that one failure cascades through the whole stack. AI workloads, especially GPU-heavy ones, run better and more cost-effectively when the infrastructure is simpler, more transparent, and built for their exact workloads. That’s why 2026 will be a major year for cloud repatriation back to regional providers and bare-metal platforms built for performance.”
– Richard Copeland, CEO of Leaseweb USA
Source: crowdfundinsider.com
