DataDome Report: Agentic Browser Traffic Surges 4.7x in September

DataDome Report: Agentic Browser Traffic Surges 4.7x in September

The rapid emergence of autonomous AI agents is fundamentally altering the landscape of web traffic, shifting the challenge from simple data scraping to complex, action-oriented interactions. According to DataDome’s Q3 2026 AI Traffic Report, AI agent traffic grew 41.5% quarter-over-quarter, reaching nearly 25 billion requests. This surge is driven largely by agentic browsers—tools that do not merely read pages but execute actions—which saw requests jump from 12.5 million to 58.2 million in September alone. This evolution signals a critical shift for enterprise security and infrastructure management.

Agentic Browser Volume and Meta Muse Impact

The surge in agentic browser activity is heavily linked to the arrival of Meta’s Muse, a tool that operates via remote browsers spun up in the cloud. Although Muse only became fully available on September 22, it generated 29.1 million requests in its first 12 days of availability. DataDome reports that Muse accounted for 33% of all agentic browser volume in September. Crucially, Muse does not identify itself as an AI agent, complicating traditional detection methods. In a specific traffic sample, over 99% of Muse requests originated from cloud and CDN infrastructure rather than individual user devices. This cloud-native behavior distinguishes it from traditional user-driven traffic, as the agent operates from its own cloud virtual machine to perform tasks on behalf of users.

Security Risks in High-Value Transactional Endpoints

Unlike standard web crawlers, agentic browsers are increasingly interacting with sensitive areas of the digital customer journey. DataDome’s analysis highlights that 8.9% of sampled Muse requests touched high-risk endpoints, including forms, logins, payments, account creation, and shopping carts. This capability to interact with transactional elements raises the stakes for enterprise security, as these agents can perform actions that mimic human intent. While Meta remains the dominant player, generating 49.7% of total AI agent requests (12.4 billion), its market share dipped slightly from 51.5% in Q2. Meanwhile, non-Meta agents are growing at a faster rate of 47%, compared to Meta's 36.5%, suggesting a diversifying ecosystem of autonomous actors.

Key Takeaways

  • Agentic browser requests increased 4.7x in September, rising from 12.5 million to 58.2 million.
  • Meta’s Muse generated 29.1 million requests within its first 12 days of full availability starting September 22.
  • 8.9% of sampled Muse requests interacted with sensitive endpoints such as login, payment, and cart functions.

TechInsyte's Take

In our view, the rise of Muse represents a paradigm shift in how enterprises must approach traffic validation. Because Muse operates from cloud infrastructure rather than residential IPs, traditional security policies that rely on distinguishing between ISP-based human traffic and data-center bots will likely fail. This signals that the "identity" of a session can no longer be assumed based on its origin. Organizations must move toward intent-based validation, where the focus is not just on where the traffic comes from, but what the agent is authorized to do once it reaches high-value endpoints like checkout flows or account management pages.

Questions & Answers

How does the behavior of agentic browsers differ from traditional web crawlers?

Unlike traditional crawlers that primarily read and index page content, agentic browsers are designed to take actions on pages, such as filling out forms or interacting with checkout processes.

Why does Meta's Muse pose a unique challenge to existing bot detection?

Muse operates from cloud and CDN infrastructure rather than user devices, and it does not identify itself as an AI agent, making it difficult for standard security protocols to distinguish it from legitimate user traffic.

What specific enterprise risks are associated with the rise in agentic traffic?

The primary risk lies in agents interacting with high-value endpoints, including login, payment, and account creation forms, which are the most frequent targets for fraud and unauthorized actions.

Is Meta the only major player in the AI agent traffic market?

No; while Meta generates the largest share at 49.7%, non-Meta agents are growing at a faster rate (47%) than Meta's agents (36.5%), indicating a diversifying landscape of autonomous tools.

Source: DataDome

TechInsyte | Technology Intelligence technology intelligence workspace

About TechInsyte | Technology Intelligence

TechInsyte is a B2B technology news and intelligence platform covering major developments across AI, cloud, cybersecurity, enterprise software, semiconductors, startups, policy, and markets. We focus on the signals that matter for decision-makers.

The idea behind TechInsyte is simple. Technology moves fast, and professionals need clear information without unnecessary noise. New platforms emerge, security risks evolve, enterprise software changes, and the AI shift continues to reshape how companies operate. We help readers understand those developments in a practical and business-focused way.

Our coverage focuses on meaningful technology updates, product launches, enterprise strategy, funding activity, regulatory change, infrastructure trends, and the broader forces shaping the technology industry. The goal is to keep every article clear, relevant, and useful for professionals who need to know what happened, why it matters, and what it could mean next.

TechInsyte is built for readers who want sharper context, cleaner coverage, and a more focused view of technology without the clutter.