The Agentic Shift: How AI Is Transforming Application Security
News | 20.11.2026
For decades, application security has been built on a single straightforward assumption: most interactions with websites and systems originate from real humans. Today, however, that assumption no longer holds. According to the Thales Bad Bot Report, automated traffic accounts for 53% of all web traffic, with malicious bots responsible for 40%. Furthermore, the number of attacks powered by artificial intelligence increased 12.5-fold year over year.
We are witnessing a fundamental shift. Applications are increasingly accessed not by people, but by intelligent software capable of independently browsing pages, making decisions, interacting with APIs, and executing complex tasks. For IT leaders and Chief Information Security Officers (CISOs), this means the very same AI technologies accelerating digital business are simultaneously changing the nature of cyberattacks.
A new era of cyber threats: from scripts to autonomous AI agents
Automation has long been part of the cybercriminal toolkit: credential stuffing, vulnerability scanning, web scraping, and account takeover. However, artificial intelligence radically increases the speed, scale, and adaptability of these operations.
Over the past year alone, Thales (Imperva) solutions blocked more than 17 trillion bad bot requests. Yet the challenge goes beyond sheer volume. Threat actors increasingly exploit legitimate application functionality, making malicious activity extremely difficult to distinguish from genuine customer behavior.
Key differences: traditional attacks vs. AI agents
Traditional automated attacks
- Toolkit: Pre-written, rigid scripts.
- Adaptability: Low — the algorithm breaks at the slightest change in website structure.
- Identification: Easily blocked using known signatures and IP addresses.
Autonomous AI attacks (Agentic AI)
- Toolkit: Self-learning algorithms and large language models (LLMs).
- Adaptability: High — AI analyzes server responses and adjusts tactics on the fly.
- Identification: Behavior closely mimics a real user or a legitimate AI assistant, severely complicating detection.
Why API security is taking center stage
Application programming interfaces have become the connective tissue of modern digital business. Unlike humans, AI agents do not need a visual website interface — they interact directly with APIs to retrieve data and execute transactions. According to statistics, 27% of all sophisticated bot attacks today specifically target APIs.
An API request can be technically valid and properly authenticated while still posing a critical threat. Attackers exploit excessive permissions, manipulate parameters, and abuse business logic (21% of attacks aim to compromise workflows). To prevent such incidents, organizations need complete visibility across their entire API infrastructure.
The future of application security lies not merely in blocking suspicious traffic, but in understanding what connections to your systems are trying to do — and ensuring they can only perform authorized actions.
How Thales (Imperva) addresses next-generation security challenges
In an environment where legitimate and malicious bots are becoming increasingly indistinguishable, and AI applications introduce new attack vectors (such as prompt injection), siloed security tools are no longer effective. The Application Security platform from Thales (Imperva) blocks over 113 billion attacks daily and delivers unified, multi-layered protection.
Web Application Firewall (WAF)
- Exploit protection: Instant blocking of critical vulnerabilities (including OWASP Top 10) and zero-day attacks.
- Accuracy: Minimizing false positives through advanced traffic analysis algorithms.
Advanced Bot Protection
- Intent recognition: Tracking business logic to detect sophisticated AI bots mimicking human behavior.
- Seamless experience: Continuous protection for websites, mobile apps, and APIs without degrading user experience.
API Security
- Deep discovery: Automated inventory of all API interfaces, including shadow and zombie endpoints.
- Data monitoring: Classification of sensitive information to reliably prevent data leaks.
AI Application Security
- AI model security: Controlling language model interactions to prevent prompt injections and unauthorized access.
- Generative AI control: Protecting internal GenAI applications integrated into corporate workflows.
Client-Side Protection
- Script control: Full visibility into third-party JavaScript code to prevent Magecart attacks and digital skimming.
- Compliance: Streamlining adherence to strict PCI DSS 4.0 standard requirements.
Expert implementation with Softprom
Building a resilient defense system requires not only cutting-edge technologies, but also the right architecture. Softprom is an official distributor of Thales (Imperva). We provide comprehensive support at every stage: from auditing your current infrastructure to platform deployment and optimization.
Our engineers possess deep expertise and will help you integrate the Application Security platform tailored to your business processes — whether in the cloud, on-premises, or in a hybrid environment.
Protect your web applications and APIs against next-generation cyber threats. Contact us to request a consultation or a free pilot project of Thales (Imperva) solutions.