How AI agents fetched /review/threatdown/ on ZeroMetric
Reporting period: Jul 6 – Jul 12, 2026 (Mon–Sun, 7 days)
ThreatDown was fetched 38 times by AI agents this week, down from 42 the prior week (-10%), with ChatGPT-User accounting for 33 of those fetches.
Activity was almost entirely confined to weekdays (38 of 38 fetches). This pattern is typical of training or indexing crawlers operating on business-hours schedules, or of B2B-driven user research. Weekend volume: 0 fetches.
| Agent | Operator | Mode | Fetches | Share |
|---|---|---|---|---|
| chatgpt-user | OpenAI | Live retrieval | 33 | 86.8% |
| anthropic | Anthropic | Training/index | 3 | 7.9% |
| oai-searchbot | OpenAI | Live retrieval | 1 | 2.6% |
| duckassistbot | DuckDuckGo | Live retrieval | 1 | 2.6% |
Methodology: every AI agent fetch of the ThreatDown review is logged by our servers and classified by its user-agent string. We distinguish live retrieval (a real user asked an AI chatbot a question, the chatbot fetched this page to answer it) from training/index (scheduled AI crawling for model updates). We publish a fresh AI Visibility report every Monday at 02:00 UTC, covering the most recent complete Mon–Sun week.
This weekly report is free and public. We're currently scoping a paid vendor product with historical trends, query-intent classification, and cross-site comparison.