Web Application Firewalls are crucial for protecting web applications against a wide range of cyber threats. Traditional Web Application Firewalls often struggle to effectively distinguish between malicious and legitimate traffic, leading to limited efficacy in threat detection. To overcome these limitations, this paper proposes an Adaptive Dual-Layer WAF employing a two-layered Machine Learning model designed to enhance the accuracy of anomaly and threat detection. The first layer employs a Decision Tree (DT) algorithm to detect anomalies by identifying traffic deviations from established normal patterns. The second layer employs Support Vector Machine to classify these anomalies as either threat anomalies or benign anomalies. Our Adaptive Dual Layer WAF incorporates comprehensive data pre-processing and feature engineering techniques and has been thoroughly evaluated using five large benchmark datasets. Evaluation using these datasets shows that ADL WAF achieves a detection accuracy of 99.88% and a precision of 100%, significantly enhancing anomaly detection and reducing false positives. These findings suggest that integrating machine learning techniques into WAFs can substantially improve web application security by providing more accurate and efficient threat detection.

As Agentic AI gain mainstream adoption, the industry invests heavily in model capabilities, achieving rapid leaps in reasoning and quality. However, these systems remain largely confined to data silos, and each new integration requires custom logic that is difficult to scale. The Model Context Protocol (MCP) addresses this challenge by defining a universal, open standard for securely connecting AI-based applications (MCP clients) to data sources (MCP servers). However, the flexibility of the MCP introduces new risks, including malicious tool servers and compromised data integrity. We present MCP Guardian, a framework that strengthens MCP-based communication with authentication, rate-limiting, logging, tracing, and Web Application Firewall (WAF) scanning. Through real-world scenarios and empirical testing, we demonstrate how MCP Guardian effectively mitigates attacks and ensures robust oversight with minimal overheads. Our approach fosters secure, scalable data access for AI assistants, underscoring the importance of a defense-in-depth approach that enables safer and more transparent innovation in AI-driven environments.

Training large-scale language models is increasingly critical in various domains, but it is hindered by frequent failures, leading to significant time and economic costs. Current failure recovery methods in cloud-based settings inadequately address the diverse and complex scenarios that arise, focusing narrowly on erasing downtime for individual tasks without considering the overall cost impact on a cluster. We introduce Unicron, a workload manager designed for efficient self-healing in large-scale language model training. Unicron optimizes the training process by minimizing failure-related costs across multiple concurrent tasks within a cluster. Its key features include in-band error detection for real-time error identification without extra overhead, a dynamic cost-aware plan generation mechanism for optimal reconfiguration, and an efficient transition strategy to reduce downtime during state changes. Deployed on a 128-GPU distributed cluster, Unicron demonstrates up to a 1.9x improvement in training efficiency over state-of-the-art methods, significantly reducing failure recovery costs and enhancing the reliability of large-scale language model training.

Abstract--Due to the increasing sophistication of web attacks, Web Application Firewalls (WAFs) have to be tested and updated regularly to resist the relentless flow of web attacks. In practice, using a brute-force attack to discover vulnerabilities is infeasible due to the wide variety of attack patterns. Thus, various black-box testing techniques have been proposed in the literature. However, these techniques suffer from low efficiency. This paper presents Reinforcement-Learning-Driven and Adaptive Testing (RAT), an automated black-box testing strategy to discover injection vulnerabilities in WAFs. In particular, we focus on SQL injection and Cross-site Scripting, which have been among the top ten vulnerabilities over the past decade. It then utilizes a reinforcement learning technique combined with a novel adaptive search algorithm to discover almost all bypassing attack patterns efficiently. We compare RAT with three state-of-the-art methods considering their objectives. The experiments show that RAT performs 33.53% and 63.16% on average better than its counterparts in discovering the most possible bypassing payloads and reducing the number of attempts before finding the first bypassing payload when testing well-configured WAFs, respectively. Thus, an enormous amount of private data of individuals and organizations is stored in web applications databases, making them tempting targets for attackers. A recent report reveals that web applications may experience up to 26 attacks per minute [1]. Moreover, according to Symantec's security report, 76% of websites are vulnerable to several attacks [2].

Pattern recognition and machine learning techniques have been increasingly adopted in adversarial settings such as spam, intrusion and malware detection, although their security against well-crafted attacks that aim to evade detection by manipulating data at test time has not yet been thoroughly assessed. While previous work has been mainly focused on devising adversary-aware classification algorithms to counter evasion attempts, only few authors have considered the impact of using reduced feature sets on classifier security against the same attacks. An interesting, preliminary result is that classifier security to evasion may be even worsened by the application of feature selection. In this paper, we provide a more detailed investigation of this aspect, shedding some light on the security properties of feature selection against evasion attacks. Inspired by previous work on adversary-aware classifiers, we propose a novel adversary-aware feature selection model that can improve classifier security against evasion attacks, by incorporating specific assumptions on the adversary's data manipulation strategy. We focus on an efficient, wrapper-based implementation of our approach, and experimentally validate its soundness on different application examples, including spam and malware detection.

Something is materially broken with web application security and, as a result, critical attacks are being missed. According to a new report from Kaspersky Lab, 73 percent of corporate network breaches in 2017 were achieved via vulnerable web applications. There are a variety of factors contributing to the problem: resource constraints, concerns over false positives and the sheer volume of the attack surface. The good news is that machine learning can be applied to help overcome these issues and strengthen application security. Most companies are failing to harness the full power of machine learning.

The Web Application Firewall (WAF) issue didn't seem to me as a big deal until I actually started to dig deeper into the ongoing discussion in this field. It generally seems that vendors are trying to convince customers and themselves that everything is going smooth and that there is not a problem. In reality, however, customers don't buy it anymore and the WAF industry is under a major pressure as constantly failing on the customer quality perspective. There have also been red flags raised from the use of the runtime application self-protection (RASP) technology. There is now a trend to enter the mitigation/defense side into the application and compile it within the code.

Today, hackers are winning the game, and the long list of successful breaches is their scorecard. Traditional endpoint security can't keep up. Outdated perimeter defenses are being rendered ineffective. And the approaches of many security vendors are only designed to fill the holes in the boat as it sinks to the bottom. Artificial intelligence (AI) and machine learning security tools, combined with human expertise, offer a better way forward.

Fortinet today at the Gartner Security & Risk Management Summit 2018 announced it has infused machine learning algorithms and user-behavioral analytics in its web application firewall to identify nearly 100 percent of all cyberthreats. John Maddison, senior vice president of products and solutions for Fortinet, said version 6.0 of the company's FortiWeb Web Application Firewall (WAF) software employs machine learning algorithms to identify both known and unknown threats. That latter capability is enabled by applying algorithms against the user behavior data being collected to identify anomalies indicative of a new, unknown threat being introduced into the IT environment. Historically, WAFs have relied on application learning (AL) to identify anomalies and known threats. But Maddison said that approach generates too many security alerts, which ultimately leads to a state of alert fatigue that makes it easy for cybersecurity professionals to miss or ignore critical information.

The assignment of weights to attacks in a classical Argumentation Framework allows to compute semantics by taking into account the different importance of each argument. We represent a Weighted Argumentation Framework by a non-binary matrix, and we characterize the basic extensions (such as w-admissible, w- stable, w-complete) by analysing sub-blocks of this matrix. Also, we show how to reduce the matrix into another one of smaller size, that is equivalent to the original one for the determination of extensions. Furthermore, we provide two algorithms that allow to build incrementally w-grounded and w-preferred extensions starting from a w-admissible extension.