PATTERN Cited by 1 source
Verified-bot delisting¶
Shape¶
A platform that operates a Verified Bots program — a
curated allowlist of bot operators whose traffic is treated as
known-good by default — uses membership as the enforcement
lever. When a member operator is found to be violating the
program's norms (running stealth
crawlers, ignoring robots.txt, rotating identity to evade
blocks), the platform delists them.
Delisting does not block the operator; it moves the operator from the allowlist into the general-population bucket where:
- Bot-management scoring applies normally.
- ML-derived stealth signatures apply normally.
- Customer-facing block rules naming the declared crawler take effect again.
- The operator no longer benefits from the reputational signal of being "verified."
Why it works¶
- Reputational leverage — being in the Verified Bots directory is a trust signal origins rely on; exit is publicly visible and damages the operator's relationships with origins and downstream customers.
- Downstream-feature exclusion — Verified Bots membership is a prerequisite for features like pay-per-crawl (requires stable identity) and for default-allow policies customers configure against the Verified Bots list. Delisting automatically withdraws those privileges.
- Ecosystem-wide effect — the 2025-08-04 Perplexity post was simultaneously a public attribution and an enforcement action: not just Cloudflare's own customers got the new block signatures, but the wider industry got a documented case study.
- Escalation asymmetry — Cloudflare operates on the origin side across ~20 % of web traffic; the operator has no equivalent cross-origin presence to retaliate from. Governance asymmetry = enforcement leverage.
Limits of the lever¶
- Single-platform scope — delisting only applies to Cloudflare's program. Other CDNs and verified-bot directories are independent.
- Doesn't stop the actual crawling — an operator that no longer cares about the reputation can continue; the follow-up enforcement is ML fingerprinting + pay-per-crawl gating, not delisting itself.
- No appeals process disclosed in the 2025-08-04 post.
Canonical instance¶
Cloudflare delists Perplexity AI from Verified Bots (2025-08-04), after confirming through a brand-new domain experiment that Perplexity's stealth crawler was accessing content on domains where their declared crawlers had been blocked. The delisting:
- Removes Perplexity from the Verified Bots directory.
- Makes Cloudflare's managed AI-bots ruleset available with stealth-crawler signatures for all customers.
- Publicly documents the violation with empirical evidence (domain-experiment design, crawler-volume numbers, UA comparison, IP + ASN rotation evidence).
Preconditions¶
- Operator must run a Verified Bots / trusted-identity program — the pattern is empty on a platform that treats all bots equivalently.
- Program must have published norms — otherwise delisting appears arbitrary.
- Platform must have empirical grounds — the 2025-08-04 post's controlled experiment + cross-customer traffic analysis is the evidentiary floor.
- Platform must be willing to absorb commercial risk — the delisted operator may be a paying customer of other platform products.
Related¶
- concepts/verified-bots — the concept anchor.
- concepts/stealth-crawler / concepts/bot-vs-human-frame.
- systems/web-bot-auth — the cryptographic-identity substrate the Verified Bots program is evolving toward.
- systems/cloudflare-bot-management / systems/pay-per-crawl.
- patterns/stealth-crawler-detection-fingerprint / patterns/stealth-on-block-fallback / patterns/signed-bot-request.
Seen in¶
- sources/2025-08-04-cloudflare-perplexity-stealth-undeclared-crawlers — canonical wiki instance.