If we think 2025 has been fast-paced, it’s going to feel like a warm-up for the changes on the horizon in 2026. Every time this year, Thales experts become cybersecurity oracles and predict where the industry is heading in the next twelve months. And, unlike the vague messages received by ancient Greeks, our predictions for 2026 are backed by our extensive experience and understanding of the cyber environment.

In the first part of this two-part series, we focus on what 2026 means for AI, quantum cryptography, and threats at all levels.

“In 2026, AI security will emerge as a formal discipline, much like application security did a decade ago.” – Nadav Avital, Senior Director, Threat Research

With three good years of AI exploration behind us, it’s time that AI security finally has its day. Organizations can no longer afford to operate in the AI space without security built in from the start. The attack surface created by AI systems introduces a new class of bespoke threats, such as prompt injection, data poisoning, model evasion, and unpredictable or rogue model behavior that can’t be effectively addressed with traditional security approaches. These challenges are distinct enough that they demand a dedicated, purpose-built security discipline of their own.

Failing to secure AI as its own entity leaves AI-driven tools vulnerable to supply-chain compromise, automated sabotage, and sensitive data leakage. While many organizations have attempted to extend existing controls or “bolt on” protections as AI evolves, this approach will not suffice. Only purpose-built, AI-focused security measures can provide the level of resilience these systems now require.

Next year, “enterprises will deploy agent-governance layers to monitor, sanitize, and sandbox AI models, enforcing identity, access, and data integrity while detecting misuse and model drift,” Avital says. Organizations that fail to do so will quickly fall behind competitors who invest early. Those that embrace these controls will not only gain an immediate security advantage but also shape the first generation of AI-defense standards, talent, and technologies that the rest of the industry will ultimately follow.

“By 2026, organizations will recognize that internal traffic is no longer inherently trusted and begin applying Zero Trust principles inside their networks. Application security will evolve beyond perimeter defense into continuous, context-aware protection within every service boundary.” – Nadav Avital

As more businesses adopt advanced agent-style AI and it becomes more embedded in internal business processes, it generates new patterns of ingress-egress API traffic and lateral system-to-system communication. Much of this activity happens behind the scenes, slipping beneath the visibility of WAFs and traditional AppSec security controls.

This shift will also force zero-trust security deeper into internal processes, causing Zero Trust Network Architecture (ZTNA) efforts to effectively double in scope and include:

• Monitoring of all API traffic
• Service-mesh-integrated WAFs
• Agent-aware analytics that can scour behavioral analytics for malicious patterns coming from compromised agents

“Attack surfaces made up of multiple cloud environments, hyper-connected systems, and thousands of dynamic entry points are creating the perfect conditions for a new class of predators to thrive as self-learning, adaptive bots that evolve with every interaction.” – Tim Chang, Vice President, Application Security

The next step in AI weaponization will be the transformation of AI agents into predator bots that can teach themselves to hunt, unleashing an abnormally powerful force against current AppSec tools.

As a result, defensive application security must shift to a more proactive stance. According to Chang, “In 2026, bot defense will shift from passive detection to active disruption to spot intent, fingerprint behavior, and intercept malicious automation before it ever reaches the application layer.”

This means that organizations are going to have to increase investments in:

• Runtime bot analytics
• Anomaly detection
• AI-against-AI countermeasures

Chang concludes that AI-powered bots will force “APIs… to finally receive the scrutiny they’ve long deserved.”

“The Imperva Threat Research team uncovered multiple high-severity zero-days in 2025, proving that even mature systems remain exposed to AI-accelerated discovery and exploitation. In 2026, the gap between disclosure and weaponization will shrink to minutes, unleashing a surge in zero-day attacks targeting application frameworks, open-source components, and APIs.” – Nadav Avital

Previously, well-established cybersecurity postures were sufficient as a defense against most low-level threats. Now, that’s not necessarily the case. AI has given low-level attackers the technological leverage they need to break down those barriers, and at a low-effort cost. With minimal skills and in record time, LLMs are now used to help attackers:

• Reverse-engineer patches
• Chain exploits
• Find logic flaws

“2026 will be a year of reckoning for suppliers and OEMs as they rush to meet the Cyber Resilience Act vulnerability management requirements. The biggest challenge won’t be the intent of the regulation, but the supply chain’s uneven readiness to comply. CISOs and product leaders will realize they’re only as compliant as their least-prepared vendor.” – Bob Burns, Chief Security Officer

If highly mature systems can still be compromised by AI-driven attacks, the risk is even greater for the uneven, developing security practices found across most supply chains. When AI-automated attacks inevitably target the weakest third-party links, non-compliance will quickly become a serious and costly problem.

The new threat and legal reality “will permanently elevate secure development lifecycle (SDL) practices from ‘best practice’ to legal obligation, reshaping how products are built, tested, and supported,” explains Bruns. “2026 is the year when security engineering becomes regulatory engineering.”

“In 2026, efficiency will become the defining metric of cyber resilience.” – Romain Deslorieux. Associate Vice President, Channel Sales, Global System Integrators

The growing need for scalable, intelligent defenses highlights another prescient trend: “resiliency through efficiency.”

Deslorieux observes that as tools are being consolidated into unified platforms, “Human expertise will shift from triage to strategy, transforming cybersecurity from a cost center into a competitive advantage built on trust and innovation.” AI enables organizations to be able to make this change, as AI-powered unification and efficiency contribute directly to the speed and scale at which teams can respond to AI-powered threats.

“Quantum computing and AI are advancing faster than most organizations can adapt. Sectors such as finance, healthcare, and critical infrastructure face the earliest deadlines, with cryptographic deprecation expected by 2030 and disallowance by 2035.” – Blair Canavan, Director, PKI & PQC Alliances 

“Quantum computing’s timeline is collapsing faster than anyone expected. Quantum readiness won’t be optional in 2026; it will be policy.” – Todd Moore, Global Vice President, Encryption 

“The quantum countdown has begun. Organizations that haven’t started planning for a post-quantum world are already behind.” – Haider Iqbal, IAM Director 

All three quotes above lead to the same conclusion: Quantum will become the new hype cycle in 2026. Not because quantum computing is new, but because we are finally approaching the inflection point at which “post-quantum readiness” moves from theoretical to existential. 

Organizations should adopt post-quantum readiness because adversaries have already begun preparing. “Even without a commercially viable quantum computer,” Iqbal says, “‘harvest-now, decrypt-later’ attacks make post-quantum authentication a present-day imperative.”

Governments, standardization bodies, and enterprises are preparing for quantum’s potential now. 

• “Governments and critical industries are already conducting [PQC] pilot programs but 2026 will be the year those pilots become requirements…” – Todd Moore
• “Standards bodies like NIST are finalizing post-quantum algorithm recommendations for public key infrastructure (PKI), setting the stage for widespread adoption.” – Bob Burns
• “By 2026, forward-leaning enterprises will pilot post-quantum authentication frameworks as part of broader crypto-agility programs. These efforts will shift from experimental labs to real-world pilots designed to safeguard identity systems before the next generation of cryptographic threats arrives.” – Haider Iqbal

While critical industries and government bodies are already conducting post-quantum pilot programs, “2026 will be the year those pilots become requirements,” Moore states. He concludes that next year, “quantum-safe migration will no longer be optional.”

This list highlights some of the forward-looking predictions from our Thales experts. Drawing on years of experience tracking security trends, they expect several meaningful shifts to emerge in 2026.

As organizations prepare for the post-quantum crossover, secure APIs against AI-driven attacks, leverage AI against AI techniques, and elevate zero-trust everywhere, they can be on the cutting edge of change.

In the next part of this series, we’ll examine what these trends mean for the business and outline practical ways organizations can get ahead of the associated risks.