Log research published by SenseGap on November 5 found that fake Googlebots can pass Google's own reverse DNS verification check. The flaw sits in step two of Google's documented four-step check, and it means a scraper running on an ordinary Google Cloud server can look exactly like Googlebot in your logs.
Google documents a four-step manual check for verifying that a request really came from one of its crawlers. First, run a reverse DNS lookup on the IP address from your logs. Second, verify that the hostname ends in googlebot.com, google.com, or googleusercontent.com. Third, run a forward DNS lookup on that hostname. Fourth, confirm the forward lookup returns the original IP address. The forward step is the part that is supposed to make the check airtight. Anyone who controls an IP block can set its reverse DNS to any name they like, including a fake crawl-1-2-3-4.googlebot.com, but only Google can make its own DNS servers answer a forward query with that address. For years, SEOs have treated a request that survives all four steps as a genuine Google crawler, and many sites wave such requests through firewalls and login walls on that basis alone.
SenseGap reported on November 5 that the four-step check can be passed from a Google Cloud server. The problem is step two: it accepts hostnames ending in googleusercontent.com. Every Google Cloud virtual machine gets a default hostname under bc.googleusercontent.com, and that hostname resolves forward to the machine's own IP address. So a request sent from a cheap cloud server passes step two because its hostname ends in googleusercontent.com, and passes step four because the forward lookup returns the same IP it came from. The check was designed to catch spoofers who fake a reverse DNS name they do not control, but a cloud server genuinely controls its own googleusercontent.com hostname, so there is nothing to catch. Google's own hostname table is stricter than the four-step summary: common crawlers such as Googlebot resolve to googlebot.com, special-case crawlers such as AdsBot resolve to google.com, and googleusercontent.com hostnames belong to user-triggered fetchers, for example when a site running on Google Cloud fetches an external feed on behalf of one of its own users. A request carrying a Googlebot user agent that resolves to a googleusercontent.com hostname is therefore not Googlebot at all, yet the simplified check waves it through.
SenseGap's research looked at requests across four sites and checked the claimed identity of each AI and search crawler against the verification method its operator publishes. The results were stark. Every request claiming to be CCBot, DuckAssistBot or Perplexity-User that could be checked turned out to be fake, and so were 32% of the requests claiming to be GPTBot. The timing fits a wider pattern: Google's Gary Illyes said in an interview reported on October 30 that bot impersonation is rising alongside AI, and that his own honeypots attract crawlers claiming to be Googlebot from IP addresses that have nothing to do with Google. The impersonators are not only scraping content. Other log analyses from the same period found single addresses presenting a dozen different AI user agents each while probing paths such as /.env and /actuator, which is somebody hunting for leaked credentials and exposed admin endpoints rather than an assistant reading a blog post.
The fix is to stop treating the four-step check as a single pass-or-fail test and to check the hostname against the right pattern for the crawler being claimed. For a request claiming to be Googlebot, accept only googlebot.com or geo.googlebot.com. For AdsBot and other special-case crawlers, accept google.com. A googleusercontent.com hostname belongs to user-triggered fetchers, not to the crawler, and should never earn Googlebot treatment. Better still, match the IP address against the JSON range lists Google publishes for its crawlers and fetchers, which removes DNS from the question entirely. The same discipline applies to AI crawlers: OpenAI, Anthropic and Perplexity each publish IP range files for their bots, and a request from an address outside the published range is not the bot it claims to be, whatever its user agent says. Bots with no published verification method cannot be verified at all, so they should be logged as unverifiable rather than counted as genuine or fake.
Fake crawlers corrupt the data SEOs make decisions from. A flood of fake Googlebot requests inflates crawl statistics, distorts log-file analysis, and can look like sudden spikes in Search Console clicks that are really bot traffic. Worse, sites that allowlist anything passing the reverse DNS check may be opening their doors to scrapers that steal content, probe for vulnerabilities, or harvest pages for AI training datasets the site owner never agreed to feed. For link builders, the lesson is that log data is only as trustworthy as the verification behind it: a page that looks heavily crawled by AI bots may simply be heavily scraped by impostors, and outreach or content decisions built on that picture will be wrong. The practical move is to re-check your own logs with the stricter hostname rules, tighten any firewall rule that trusts googleusercontent.com as Googlebot, and treat every unverifiable crawler claim as unproven until its IP address says otherwise.
Sources: SenseGap, November 5, 2026.
Related reading on this site: Google: Technical SEO Remains Key for AI SEO, Google Explains How Redirects Affect Ranking Signals, November Search Console Click Spikes Are Bot Traffic, Google Rank Data Issues, Lighthouse Update and AI Opt-Out, Brave Sells Its Search Index to AI — Your Site Is Probably In It, and Google Loses Again to SerpApi in Scraping Case.
Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.
Start free trial