SYSTEM Cited by 1 source

Perplexity AI

Perplexity AI is an AI-powered answer engine (product site perplexity.ai) that answers user queries by fetching web content at query time and summarizing it with an LLM. Stub page on the wiki because Perplexity's retrieval infrastructure is the subject of Cloudflare's 2025-08-04 investigation.

Retrieval infrastructure as documented by Cloudflare (2025-08-04)

Perplexity documents two declared crawlers:

• PerplexityBot — declared training / indexing crawler.
• Perplexity-User — declared on-demand user-triggered fetcher.

Both declared crawlers respect robots.txt and use documented IP ranges / user agents.

Cloudflare's 2025-08-04 post documents a third, undeclared crawler that:

• Impersonates a generic Chrome/124.0.0.0 user agent on macOS.
• Sources requests from IPs outside Perplexity's published IP range.
• Rotates across multiple ASNs.
• Does not fetch or respect robots.txt.
• Activates when declared crawlers are blocked.

Cloudflare de-listed Perplexity from its Verified Bots program and shipped ML-based block signatures for the stealth crawler into its managed AI-crawler ruleset as a result. See patterns/verified-bot-delisting and patterns/stealth-on-block-fallback.

Seen in

• sources/2025-08-04-cloudflare-perplexity-stealth-undeclared-crawlers — canonical wiki instance; enumerates the two declared crawlers, documents the stealth third crawler, and is the delisting event.

Related

• systems/chatgpt-user — positive-control counterpart in the same Cloudflare post.
• systems/web-bot-auth — cooperative-crawler identity scheme Perplexity is not a member of (post-delisting).
• concepts/stealth-crawler / concepts/undeclared-crawler.
• companies/cloudflare.