EN
FR
ES
DE
IT
KO
JA
An AI bot signature is a user-agent pattern (and sometimes an IP range) that uniquely identifies a generative AI crawler — OpenAI’s GPTBot, Anthropic’s ClaudeBot, PerplexityBot, Google-Extended, and so on. AI Labs Audit maintains a registry of 256 active signatures, continuously updated.
Why a dedicated registry?
AI bots evolve quickly: new entrants (Mistral-AI-User, ByteDance-Sora), renamings (Anthropic-AI → ClaudeBot → Claude-Web → Claude-SearchBot), behaviour shifts (training vs. search mode). Without an up-to-date registry, your robots.txt and WAF risk either blocking the right bots or ignoring the ones actually consuming your content.
What the AI Labs Audit registry contains
For each signature we track:
- The publisher (OpenAI, Anthropic, Google, Microsoft, Perplexity, xAI, ByteDance, Common Crawl…);
- The role (training, search, agent, browse) — a single publisher may run multiple bots with different roles;
- The user-agent pattern to match;
- Official IP ranges (when published);
- robots.txt recommendations (allow / disallow / contextual);
- Impact on AI visibility if the bot is blocked.
In total, 256 signatures are currently active in the registry.
Tracking-side usage
AI Labs Audit’s open-source tracker uses this registry to classify every hit received on your site. You get: most active bots, most crawled URLs per bot, crawl ↔ citation correlation across your audits, alerts on suspicious spikes or missing bots.
GEO-audit-side usage
An audit’s GEO score penalises sites that block AI search bots (PerplexityBot, OAI-SearchBot, Claude-SearchBot) because they are essential for appearing in web-augmented AI answers. Conversely, blocking training bots (GPTBot training, Google-Extended training) is neutral and even valued for brands wanting to protect their content.
Every question asked to ChatGPT without your client's name in the answer is a competitor recommended instead of them — measured across 6,820 real AI answers.