November 2025 Summaries

The emergence of Agentic AI, characterized by autonomous agents interacting with content in real time, marks a shift in the AI landscape, as detailed in the Global Bot Security Report. Traditional methods like "pay-per-crawl" are becoming obsolete, leading to the development of the Know Your Agent (KYA+Pay) model, which allows businesses to identify, verify, and monetize AI interactions securely. Partnering with Skyfire, DataDome offers a comprehensive solution that combines visibility, control, and monetization of AI traffic, enabling businesses to establish trust-based interactions with AI agents. This approach provides flexible monetization strategies, enhances security against unauthorized access, and ensures seamless integration across various infrastructures. As the agent economy evolves, this model allows companies to transform AI interactions into a reliable revenue source while maintaining robust control and visibility over their content.

The Model Context Protocol (MCP) has rapidly become the universal standard for integrating AI systems with external tools and data, embraced by major tech giants like Microsoft, Google, and OpenAI. However, its swift adoption has introduced significant security challenges, as researchers have identified extensive command injection vulnerabilities across numerous MCP servers, leading to real-world attacks involving credential theft and remote code execution. The protocol creates new attack surfaces, allowing AI agents to execute unauthorized commands or exfiltrate data if MCP servers are inadequately secured. Additionally, the storage of OAuth tokens on MCP servers elevates enterprise-wide risks, as a single breach could compromise access to multiple connected platforms. As attacks have already been carried out, safeguarding MCP implementations requires a comprehensive security strategy, including robust authentication, input validation, runtime monitoring, and human oversight for high-risk activities.

AI shopping agents are revolutionizing online purchasing by automating tasks such as browsing, price comparison, and completing transactions, leading to a significant increase in e-commerce traffic, as evidenced by DataDome's detection of nearly 1.2 billion requests from OpenAI crawlers in June 2025. This trend could potentially drive the US B2C retail market to generate up to $1 trillion in orchestrated revenue and globally between $3 trillion to $5 trillion, as per McKinsey research. However, the rise of agentic AI introduces new security challenges, as traditional security models based on static classifications are insufficient for these adaptable and autonomous agents. Modern security measures must focus on intent-based detection and trust-based security, differentiating access levels according to agent identity, behavior, and business relationships, to ensure legitimate AI agents are not blocked, thus preserving sales opportunities. Additionally, the introduction of Model Context Protocol (MCP) servers, which standardize AI agents' access to systems, necessitates robust authentication and behavior monitoring to prevent new attack vectors, emphasizing the need for businesses to develop adaptive security strategies as AI agent capabilities rapidly evolve.

Retail is undergoing a significant transformation with the advent of agentic commerce, where autonomous digital agents can discover products, compare offers, and complete purchases, bringing both opportunities and risks. These agents can drive growth and revenue but also pose threats like fraud and data loss, necessitating that online retailers adapt their platforms to accommodate AI agents securely. As highlighted by Walmart's and Salesforce's recent initiatives, the ability to manage and authenticate these AI agents will determine the success of businesses in this new landscape, as those failing to do so risk losing control over analytics, pricing intelligence, and demand signals. The 2025 Global Bot Security Report reveals that most retail and other digital transaction sectors are poorly prepared to detect even basic bots, emphasizing the need for robust bot and agent management as foundational infrastructure. Companies that can effectively differentiate between human and AI traffic while ensuring data integrity stand to gain competitive advantages, with solutions like DataDome offering tools to enhance visibility, control, and trust in this rapidly evolving digital economy.

AI agents are increasingly present on websites, serving roles ranging from helpful shopping assistants to malicious content scrapers, necessitating a shift from traditional bot detection methods to a more nuanced approach known as agentic trust management. This approach evaluates AI agents based on their intent and behavior rather than a binary human-or-bot classification, recognizing that AI agents can adapt and change behavior mid-session. The need for continuous monitoring and adaptive authentication policies is emphasized, as legitimate AI agents can become compromised over time. McKinsey research suggests significant economic potential from agentic commerce, with the US B2C retail market alone potentially generating up to $1 trillion in orchestrated revenue. Comprehensive visibility is crucial for organizations before implementing trust management policies, enabling them to safely engage with the agent economy by allowing beneficial automation and maintaining robust security.