-
Anthropic Mythos and Project Glasswing: What Frontier AI Means for Defensive Cybersecurity
Cybersecurity has historically been constrained by one scarce resource: expert attention. Modern software contains millions of lines of code, enormous dependency trees and constantly changing infrastructure, while experienced vulnerability researchers can deeply investigate only a fraction of that attack surface. Frontier AI is beginning to change that equation. Anthropic’s Claude Mythos line represents one of…
-
AI-Assisted Penetration Testing: Where Automation Helps and Where Human Researchers Matter
Penetration testing has always combined tools with human judgment. Security scanners discover exposed services, automated crawlers map applications, static analyzers inspect code and specialized utilities help researchers reproduce suspicious behavior. What is changing in 2026 is the amount of reasoning that can now be delegated to artificial intelligence. Modern AI systems can help researchers understand…
-
How AI Is Changing Vulnerability Discovery and Offensive Security
Vulnerability research has traditionally been a highly manual discipline. Security researchers read source code, reverse-engineer application behavior, compare assumptions with implementation details, reproduce suspicious conditions and gradually build an understanding of how a weakness might become exploitable. Artificial intelligence is beginning to change that workflow. Modern frontier models can reason across large codebases, analyze unfamiliar…
-
Adversarial Machine Learning: Evasion, Poisoning, Privacy and Misuse Attacks Explained
Machine learning systems are increasingly responsible for decisions that used to be handled entirely by deterministic software. Models classify information, detect fraud, identify malicious activity, recommend actions, generate content and increasingly interact with software through AI agents. As these systems become more important, attackers have an incentive to manipulate not only the applications surrounding the…
-
How to Secure AI Tools and Function Calling Against Abuse
AI applications become significantly more powerful when a model can do more than generate text. Tool use and function calling allow an LLM or AI agent to query databases, call APIs, retrieve files, send messages, create objects and trigger application logic. In MCP, for example, tools are explicitly defined as functions that models can actively…
-
MCP Security: How to Secure Model Context Protocol Servers, Tools and Authorization
MCP Security: How to Model Context Protocol has quickly become an important integration layer for AI applications and autonomous agents. Instead of building a custom connection every time an AI system needs to reach a database, development environment, business application or external service, MCP provides a standardized way for AI applications to discover data and…
-
RAG Security: Protecting Retrieval-Augmented Generation Systems From Data and Prompt Attacks
Retrieval-Augmented Generation has become one of the most practical ways to connect large language models with private or frequently changing information. Instead of relying only on what a model learned during training, a RAG application searches an external knowledge source, retrieves information relevant to the current request and places that information into the model’s context…
-
Excessive Agency in AI Systems: When Too Much Autonomy Becomes a Security Risk
Artificial intelligence becomes more useful as it gains the ability to act. A simple language model can answer a question. An AI agent can search for information, choose tools, interact with APIs, change records, send messages and continue working across several steps without waiting for a new instruction from the user. That additional autonomy creates…
-
AI Agent Security: How to Control Tools, Permissions and Autonomous Actions
AI agents represent one of the most important shifts in modern application security. A traditional chatbot receives a request and generates a response. An AI agent can go further: it can retrieve information, select tools, call APIs, modify records, interact with files, trigger external services and continue executing multiple steps without requiring a new instruction…
-
AI Red Teaming vs Traditional Penetration Testing: What Changes?
Artificial intelligence is changing the way modern applications behave, but it is also changing the way those applications need to be tested. A traditional penetration test is designed to identify exploitable weaknesses in software, infrastructure, authentication, APIs and business logic. AI red teaming expands that objective by asking a different class of questions: can an…
