What Cloudflare's June 2026 AI Audit Logs for Bots Mean for Webflow Publishers

Why Did Cloudflare Suddenly Start Logging AI Bot Visits to My Webflow Sites in June 2026?

I logged into the Cloudflare dashboard for one of my Webflow client sites on a Thursday morning in mid June 2026 and noticed a new menu item under the bot management section called AI Audit Logs. The product had not existed the day before. I clicked through and saw a per-visit log of every request from a known AI crawler over the past seven days. Forty-three percent of the bot traffic on the client's site was from AI crawlers I did not even know existed.

The Cloudflare AI Audit Logs feature shipped on June 17, 2026, as part of the company's broader response to growing publisher pressure around AI training data. The product gives a Webflow publisher granular visibility into which AI bots crawled which URLs, when, and what they pulled. For studios that manage client Webflow sites, this is the most concrete tool we have ever had to answer the publisher question of who is using my content.

This article walks through what the new logs actually contain, why the launch matters for Webflow publishers specifically, what the data has already taught me about three client sites I run, and how I would use it inside your studio this week.

What Are AI Audit Logs and Why Do They Matter for Webflow Publishers?

AI Audit Logs are a per-request log of every AI bot that hit a Cloudflare-protected site, with the bot name, the timestamp, the URL path, the response status, and the user agent string. Cloudflare exposes the logs in the dashboard and through the Cloudflare Logpush API for sites on the Pro plan and above. The feature is free at the Pro tier and bundled with Argo at the Business tier.

For Webflow publishers, this matters because Webflow does not give you origin-level logs of who crawled your site. The Webflow dashboard shows you visitor analytics, not bot logs. If you front your Webflow site with Cloudflare, the AI Audit Logs give you the visibility Webflow itself does not. According to Cloudflare's launch post on June 17, 2026, the company logged over 70 billion AI bot requests across its network in the prior month, an 11x increase compared to the same month in 2025.

For Webflow studios running client retainers, this changes the reporting conversation. Instead of telling a client we think their content is being read by AI, you can show them which AI bots came to the site, on which day, and what they pulled. That is the kind of evidence that changes how publishers think about AI training opt-outs.

Which AI Bots Show Up Most in the Logs for a Typical Webflow Site?

Across the three Webflow client sites I checked, the top five AI bots by request volume in the past seven days were GPTBot from OpenAI, Claude-Web from Anthropic, PerplexityBot from Perplexity, Google-Extended from Google, and Amazonbot from Amazon. Each of those bots has a different stated purpose, and Cloudflare labels them accordingly in the logs.

GPTBot identifies itself as the OpenAI search and training crawler. Google-Extended is the AI training opt-out signal that Google added in 2023. Claude-Web is the Anthropic web search crawler. PerplexityBot is the Perplexity citation crawler, which feeds the answer engine. Amazonbot's purpose is less clear, but it has grown 380 percent year over year according to the Cloudflare June 2026 traffic report.

The mix matters because the bots have different policies for how they treat your robots.txt and your llms.txt. PerplexityBot, in particular, has had a public dispute with publishers over whether it respects robots.txt at all. The audit logs let you check the claim instead of taking the bot operator's word.

How Do I Set Up AI Audit Logs in Front of a Webflow Site?

The setup is two steps if your Webflow site already runs through Cloudflare. Step one, enable AI Audit Logs in the Cloudflare dashboard under bot management. The toggle is on by default for new sites added after June 17, 2026, but older sites need a manual opt-in. Step two, configure how long you want to retain the logs. The Pro plan retains 30 days. The Business plan retains 90. Enterprise retains a year.

If your Webflow site does not run through Cloudflare yet, the prerequisite is a DNS-only or proxied Cloudflare zone in front of the Webflow custom domain. Webflow's edge handles SSL and CDN, but Cloudflare can sit in front for bot management. I walked through the cost trade-off in my piece on the Cloudflare free tier bandwidth cap on Webflow hosting. For most Webflow brochure sites under five million monthly page views, the Pro plan at 25 US dollars per zone is the right starting point.

The Webflow site itself does not need any changes. The logging happens at the network layer, before the request hits the Webflow origin.

But What About Bots That Hide Their User Agent?

This is the legitimate concern about any user agent-based logging. A determined scraper can change the user agent string to look like a regular browser. Cloudflare's AI Audit Logs do not catch those bots in the AI category. They show up in the regular bot management score instead, which Cloudflare assigns based on behavior signals like request patterns, JA4 fingerprints, and IP reputation.

According to the Cloudflare Radar 2025 year-end report, roughly 14 percent of automated traffic to publisher sites uses a forged user agent. That traffic is invisible to the AI Audit Logs but visible to the bot score. For a complete picture, you need both. Cloudflare's documentation recommends pairing AI Audit Logs with bot management rules that challenge low-score traffic.

For Webflow publishers worried about scraping by AI training pipelines that hide their identity, the answer is not the audit logs alone. It is the bot score paired with a managed challenge rule that catches non-AI declared bots that exhibit AI-like crawl patterns.

What Did the Logs Tell Me About Three of My Client Webflow Sites?

The first client runs a B2B SaaS marketing site. AI bots accounted for 41 percent of total bot traffic in the past seven days. GPTBot led by a wide margin at 19 percent, followed by Claude-Web at 8 percent. The pattern of URL hits was concentrated on the pricing page and the integration directory, which lines up with what AI shopping assistants pull when a buyer asks about software options.

The second client runs an Indian fintech blog. AI bot share was 67 percent of bot traffic, which surprised me. PerplexityBot was the largest at 22 percent. The pattern of URL hits was heavy on long-form explainer posts, the kind of content that answers a finance question. That tells me AI search is becoming a major distribution channel for finance content in India.

The third client is a portfolio site for a freelance designer. AI bot share was only 9 percent. The pattern of hits was concentrated on the case study pages, not the about page or the contact page. AI bots care about portfolio work as evidence, not about the personal narrative. That insight changed how I structure case studies on portfolio sites for clients going forward.

How Does This Change How I Report to Clients on Retainers?

My client reports used to show Google Search Console traffic, GA4 sessions, and the standard SEO health checks. Starting in late June 2026, I added a fourth section on AI visibility, populated from the Cloudflare AI Audit Logs. The section answers three questions for the client. Which AI bots crawled the site this month, which URLs they hit most, and whether the bot traffic is growing or shrinking compared to the prior month.

This pairs naturally with the AI citation reports I wrote about in my piece on switching client reports from PDFs to live AI visibility dashboards. Together, the two reports answer both halves of the AI visibility question. Who is crawling the content, and who is citing it back. According to a Search Engine Land study from May 2026, AI citation share grew 240 percent year over year for B2B marketing sites.

For clients in industries where AI training matters legally, the audit logs also become evidence for an opt-out conversation. A retainer client in healthtech used the logs to support a robots.txt update that explicitly blocks ClaudeBot and GPTBot from a specific subdirectory of confidential whitepapers.

How Do You Know If You Should Block, Allow, or Monitor AI Bots on a Webflow Site?

The choice depends on what you are publishing. For marketing content where discoverability matters, allow the bots. For content that exists to drive AI citations of your brand, allow the bots and add structured data to make the citations easier. For paywalled content, premium research, or anything you license commercially, block the bots that train on your content while allowing the ones that cite it back.

The distinction between training and citation is the one that drives the policy. GPTBot, Claude-Web, and Google-Extended are training crawlers. PerplexityBot, ChatGPT-User, and Claude-User are citation crawlers that feed answer engines. Most publishers I work with are happy to be cited and reluctant to be trained on. Cloudflare's bot rules let you set different policies per bot, which is what makes this manageable in 2026.

For broader context on how publishers are thinking about the AI training and citation split, my notes on the June 2026 W3C proposal to standardize llms.txt cover the standards side of this conversation.

How Should You Use Cloudflare AI Audit Logs on a Webflow Site This Week?

The three-day plan is the same one I rolled out for two clients last week. Day one, enable AI Audit Logs in Cloudflare for the Webflow site and pick the retention window. Day two, pull the past seven days of logs and identify the top five AI bots by request volume, plus the top ten URLs they hit. Day three, write a short paragraph for the client report that summarizes the AI bot activity, with a clear next-action recommendation for any URLs that should be blocked or allowed differently.

The whole exercise takes under three hours per site for a first run, and the upkeep after that is a monthly check during the regular reporting cycle. For Webflow publishers building toward stronger AI visibility, my piece on building a Webflow author archive page that earns AI citations covers the on-page work that pairs with the network-level visibility this article describes.

If you want help setting up AI Audit Logs in front of a Webflow site and turning the data into a client-ready report, I am happy to walk through it. Let's chat.