{"id":1872841,"date":"2026-09-15T14:05:33","date_gmt":"2026-09-15T12:05:33","guid":{"rendered":"https:\/\/blog.wedos.com\/?p=1872841"},"modified":"2026-09-15T14:10:22","modified_gmt":"2026-09-15T12:10:22","slug":"ai-roboti-se-meni-a-wedos-global-protection-s-nimi","status":"publish","type":"post","link":"https:\/\/blog.wedos.com\/cs\/ai-roboti-se-meni-a-wedos-global-protection-s-nimi","title":{"rendered":"AI roboti se m\u011bn\u00ed. A WEDOS Global Protection s nimi"},"content":{"rendered":"\n

AI slu\u017eby dnes nepos\u00edlaj\u00ed na weby jen jeden typ robota. Stejn\u00fd poskytovatel m\u016f\u017ee provozovat crawler pro z\u00edsk\u00e1v\u00e1n\u00ed dat, robota pro AI vyhled\u00e1v\u00e1n\u00ed i klienta, kter\u00fd str\u00e1nku otev\u0159e na p\u0159\u00edm\u00fd po\u017eadavek u\u017eivatele. A pr\u00e1v\u011b tady se internet za\u010d\u00edn\u00e1 m\u011bnit.<\/p>\n\n\n\n\n\n\n\n

D\u0159\u00edve bylo relativn\u011b jednoduch\u00e9 rozli\u0161it \u010dlov\u011bka, klasick\u00fd vyhled\u00e1va\u010d a \u0161kodliv\u00e9ho robota. Dnes mezi nimi vznik\u00e1 nov\u00e1 a rychle rostouc\u00ed skupina provozu – AI roboti.<\/strong><\/p>\n\n\n\n

A stejn\u011b rychle, jako se m\u011bn\u00ed oni, se mus\u00ed m\u011bnit i WEDOS Global Protection.<\/p>\n\n\n\n

V posledn\u00ed dob\u011b jsme proto ud\u011blali dal\u0161\u00ed v\u011bt\u0161\u00ed krok a za\u010dali zn\u00e1m\u00e9 AI roboty nejen identifikovat, ale tak\u00e9 rozd\u011blovat podle toho, co skute\u010dn\u011b d\u011blaj\u00ed<\/strong>.<\/p>\n\n\n\n

GPTBot nen\u00ed ChatGPT-User<\/h2>\n\n\n\n

Dobr\u00fdm p\u0159\u00edkladem je OpenAI.<\/p>\n\n\n\n

Pod jednou spole\u010dnost\u00ed m\u016f\u017ee p\u0159ij\u00edt n\u011bkolik velmi odli\u0161n\u00fdch typ\u016f provozu.<\/p>\n\n\n\n

GPTBot<\/strong> m\u016f\u017ee systematicky proch\u00e1zet velk\u00e9 mno\u017estv\u00ed obsahu.<\/p>\n\n\n\n

OAI-SearchBot<\/strong> z\u00edsk\u00e1v\u00e1 data pro AI vyhled\u00e1v\u00e1n\u00ed.<\/p>\n\n\n\n

ChatGPT-User<\/strong> m\u016f\u017ee p\u0159ij\u00edt ve chv\u00edli, kdy konkr\u00e9tn\u00ed u\u017eivatel po\u017e\u00e1d\u00e1 ChatGPT o otev\u0159en\u00ed ur\u010dit\u00e9 str\u00e1nky.<\/p>\n\n\n\n

Z pohledu serveru jde poka\u017ed\u00e9 o automatizovan\u00e9 HTTP po\u017eadavky.<\/p>\n\n\n\n

Jejich v\u00fdznam je ale \u00fapln\u011b jin\u00fd.<\/p>\n\n\n\n

Proto dnes AI provoz rozd\u011blujeme nap\u0159\u00edklad do t\u011bchto skupin:<\/p>\n\n\n\n

Typ<\/th>Co d\u011bl\u00e1<\/th><\/tr><\/thead>
training<\/code><\/td>systematicky z\u00edsk\u00e1v\u00e1 obsah pro pr\u00e1ci s AI modely<\/td><\/tr>
search_index<\/code><\/td>proch\u00e1z\u00ed web pro AI vyhled\u00e1v\u00e1n\u00ed a tvorbu odpov\u011bd\u00ed<\/td><\/tr>
on_demand<\/code><\/td>po\u017eadavek vznik\u00e1 p\u0159\u00edmo akc\u00ed u\u017eivatele AI asistenta<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
\"\"\n\t\t\t\n\t\t\t\t\n\t\t\t<\/svg>\n\t\t<\/button>
Po\u010det p\u0159\u00edstup\u016f z AI syst\u00e9mu v pr\u016fm\u011brnou ned\u011bli.<\/figcaption><\/figure>\n\n\n\n

Tohle rozd\u011blen\u00ed nen\u00ed jen pro statistiky. Umo\u017e\u0148uje n\u00e1m s jednotliv\u00fdmi druhy provozu zach\u00e1zet rozd\u00edln\u011b.<\/p>\n\n\n\n

Data jsou cenn\u00e1. A ka\u017ed\u00fd je chce co nejrychleji<\/h2>\n\n\n\n

Provozovatel\u00e9 AI syst\u00e9m\u016f pot\u0159ebuj\u00ed data. Hodn\u011b dat. A ide\u00e1ln\u011b co nejrychleji.<\/p>\n\n\n\n

Crawler se proto nemus\u00ed chovat jako klasick\u00fd n\u00e1v\u0161t\u011bvn\u00edk, kter\u00fd otev\u0159e str\u00e1nku, chv\u00edli ji \u010dte a potom klikne d\u00e1l. M\u016f\u017ee b\u011bhem kr\u00e1tk\u00e9 doby proj\u00edt stovky nebo tis\u00edce URL.<\/p>\n\n\n\n

Samo o sob\u011b to samoz\u0159ejm\u011b neznamen\u00e1 \u00fatok. Z pohledu serveru ale m\u016f\u017ee b\u00fdt v\u00fdsledek velmi podobn\u00fd. I legitimn\u00ed robot dok\u00e1\u017ee vytvo\u0159it takovou z\u00e1t\u011b\u017e, \u017ee za\u010dne ovliv\u0148ovat b\u011b\u017en\u00e9 n\u00e1v\u0161t\u011bvn\u00edky webu. A s podobn\u00fdmi p\u0159\u00edpady jsme se u\u017e v minulosti setkali.<\/p>\n\n\n\n

Proto nechceme zn\u00e1m\u00e9 AI roboty jednodu\u0161e ozna\u010dit jako d\u016fv\u011bryhodn\u00e9 a pustit je bez omezen\u00ed.<\/p>\n\n\n\n

Pot\u0159ebujeme v\u011bd\u011bt, kdo jsou, co d\u011blaj\u00ed a v p\u0159\u00edpad\u011b pot\u0159eby jejich provoz omezit d\u0159\u00edv, ne\u017e za\u010dne z\u00e1kazn\u00edkovi vadit<\/strong>.<\/p>\n\n\n\n

Jeden po\u017eadavek u\u017eivatele nemus\u00ed znamenat jeden request<\/h2>\n\n\n\n

Je\u0161t\u011b zaj\u00edmav\u011bj\u0161\u00ed jsou po\u017eadavky vyvolan\u00e9 p\u0159\u00edmo u\u017eivateli AI slu\u017eeb. U\u017eivatel nap\u0159\u00edklad polo\u017e\u00ed ot\u00e1zku a o\u010dek\u00e1v\u00e1 odpov\u011b\u010f b\u011bhem n\u011bkolika sekund. AI slu\u017eba kv\u016fli tomu m\u016f\u017ee pot\u0159ebovat z\u00edskat informace z n\u011bkolika str\u00e1nek nebo z jedn\u00e9 str\u00e1nky st\u00e1hnout v\u00edce zdroj\u016f sou\u010dasn\u011b.<\/p>\n\n\n\n

Jeden po\u017eadavek u\u017eivatele tak nemus\u00ed znamenat jeden HTTP request. M\u016f\u017ee jich vzniknout n\u011bkolik, n\u011bkdy i des\u00edtky t\u00e9m\u011b\u0159 sou\u010dasn\u011b.<\/p>\n\n\n\n

Z\u00e1le\u017e\u00ed tak\u00e9 na tom, jak m\u00e1 konkr\u00e9tn\u00ed AI slu\u017eba navr\u017eenou cache. Ne ka\u017ed\u00fd AI syst\u00e9m si jednou z\u00edskan\u00fd obsah ulo\u017e\u00ed a pou\u017eije ho p\u0159i dal\u0161\u00edm dotazu. \u010c\u00e1st provozu tak m\u016f\u017ee vznikat opakovan\u011b.<\/p>\n\n\n\n

A u\u017eivatel na druh\u00e9 stran\u011b samoz\u0159ejm\u011b \u010dekat nechce. V\u00fdsledkem je tlak na co nejrychlej\u0161\u00ed paraleln\u00ed z\u00edsk\u00e1v\u00e1n\u00ed dat. Z pohledu AI slu\u017eby logick\u00e9. Z pohledu webserveru u\u017e to ale m\u016f\u017ee vypadat jako velmi agresivn\u00ed automatizovan\u00fd provoz.<\/p>\n\n\n\n

Zn\u00e1m\u00fd AI robot neznamen\u00e1 neomezen\u00fd p\u0159\u00edstup<\/h2>\n\n\n\n

Proto pro n\u00e1s informace:<\/p>\n\n\n\n

tato IP pat\u0159\u00ed zn\u00e1m\u00e9mu AI providerovi\n<\/code><\/pre>\n\n\n\n
neznamen\u00e1:<\/p>\n\n\n\n
obejdi v\u0161echny ochrany\n<\/code><\/pre>\n\n\n\n
Znamen\u00e1 pouze, \u017ee v\u00edme v\u00edce o p\u016fvodu provozu.<\/p>\n\n\n\n
Cel\u00fd princip se d\u00e1 zjednodu\u0161it na t\u0159i kroky:<\/p>\n\n\n\n
ov\u011b\u0159en\u00ed \u2192 klasifikace \u2192 akce\n<\/code><\/pre>\n\n\n\n
Nejd\u0159\u00edv chceme v\u011bd\u011bt, kdo po\u017eadavek pos\u00edl\u00e1. Potom, k \u010demu dan\u00fd robot slou\u017e\u00ed. A teprve nakonec se rozhoduje, jak s n\u00edm zach\u00e1zet. V\u00fdsledkem m\u016f\u017ee b\u00fdt nap\u0159\u00edklad b\u011b\u017en\u00e9 povolen\u00ed provozu, rate limit, ov\u011b\u0159en\u00ed pomoc\u00ed Proof of Work nebo \u00fapln\u00e9 zablokov\u00e1n\u00ed.<\/p>\n\n\n\n
D\u016fle\u017eit\u00e9 je, \u017ee identita robota a jeho opr\u00e1vn\u011bn\u00ed nejsou tot\u00e9\u017e<\/strong>.<\/p>\n\n\n\n
Pro\u010d je rozd\u011blen\u00ed AI robot\u016f d\u016fle\u017eit\u00e9<\/h2>\n\n\n\n
Systematick\u00fd crawler pro z\u00edsk\u00e1v\u00e1n\u00ed velk\u00e9ho mno\u017estv\u00ed obsahu m\u00e1 jin\u00fd charakter ne\u017e po\u017eadavek vytvo\u0159en\u00fd konkr\u00e9tn\u00edm u\u017eivatelem AI asistenta. Pokud crawler za\u010dne stahovat p\u0159\u00edli\u0161 agresivn\u011b, m\u016f\u017eeme jeho provoz omezit.<\/p>\n\n\n\n
U on_demand<\/code> provozu naopak v\u00edme, \u017ee za requestem m\u016f\u017ee \u010dekat skute\u010dn\u00fd \u010dlov\u011bk, kter\u00fd pr\u00e1v\u011b pou\u017e\u00edv\u00e1 AI slu\u017ebu a o\u010dek\u00e1v\u00e1 odpov\u011b\u010f. Pr\u00e1v\u011b proto nechceme v\u0161echny AI roboty nah\u00e1zet do jednoho seznamu a aplikovat na n\u011b stejn\u00e9 pravidlo.<\/p>\n\n\n\n
Pot\u0159ebujeme m\u00edt mo\u017enost \u0159\u00edct:<\/p>\n\n\n\n
zn\u00e1me t\u011b\nv\u00edme, co d\u011bl\u00e1\u0161\nale pokud to p\u0159e\u017eene\u0161, zpomal\u00edme t\u011b\n<\/code><\/pre>\n\n\n\n
To je z\u00e1sadn\u00ed rozd\u00edl proti klasick\u00e9mu whitelistu.<\/p>\n\n\n\n
AI provoz bude d\u00e1l r\u016fst<\/h2>\n\n\n\n
Dnes \u0159e\u0161\u00edme crawlery, AI vyhled\u00e1v\u00e1n\u00ed a po\u017eadavky vyvolan\u00e9 u\u017eivatelem.<\/p>\n\n\n\n
Dal\u0161\u00edm krokem budou st\u00e1le \u010dast\u011bji AI agenti, kte\u0159\u00ed budou za u\u017eivatele proch\u00e1zet weby, porovn\u00e1vat nab\u00eddky, z\u00edsk\u00e1vat informace nebo p\u0159\u00edmo komunikovat s webov\u00fdmi slu\u017ebami.<\/p>\n\n\n\n
Rozd\u011blen\u00ed na:<\/p>\n\n\n\n
\u010dlov\u011bk \/ robot\n<\/code><\/pre>\n\n\n\n
u\u017e proto dlouhodob\u011b nebude sta\u010dit. A brzy nebude sta\u010dit ani:<\/p>\n\n\n\n
dobr\u00fd robot \/ \u0161patn\u00fd robot\n<\/code><\/pre>\n\n\n\n
Pot\u0159ebujeme v\u011bd\u011bt, kdo p\u0159ich\u00e1z\u00ed, pro\u010d p\u0159ich\u00e1z\u00ed, jak se chov\u00e1 a jak velkou z\u00e1t\u011b\u017e vytv\u00e1\u0159\u00ed<\/strong>. Pr\u00e1v\u011b t\u00edmto sm\u011brem te\u010f posouv\u00e1me WEDOS Global Protection. AI roboti nejsou automaticky probl\u00e9m. Ale nejsou ani provoz, kter\u00e9mu bychom m\u011bli bez dal\u0161\u00edho otev\u0159\u00edt dve\u0159e doko\u0159\u00e1n. <\/p>\n\n\n\n
A \u010d\u00edm v\u011bt\u0161\u00ed \u010d\u00e1st internetu budou AI syst\u00e9my proch\u00e1zet, t\u00edm d\u016fle\u017eit\u011bj\u0161\u00ed bude ten rozd\u00edl um\u011bt poznat.<\/p>\n","protected":false},"excerpt":{"rendered":"
AI slu\u017eby dnes nepos\u00edlaj\u00ed na weby jen jeden typ robota. Stejn\u00fd poskytovatel m\u016f\u017ee provozovat crawler pro z\u00edsk\u00e1v\u00e1n\u00ed dat, robota pro AI vyhled\u00e1v\u00e1n\u00ed i klienta, kter\u00fd str\u00e1nku otev\u0159e na p\u0159\u00edm\u00fd po\u017eadavek u\u017eivatele. A pr\u00e1v\u011b tady se internet za\u010d\u00edn\u00e1 m\u011bnit.<\/p>\n","protected":false},"author":2,"featured_media":1872956,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[9],"tags":[525,536,524,554,527,548,532,541,555,540,547,522,537,542,529,43,531,557,530,545,492,546,504,539,552,523,543,538,533,544,556,528,535,534,526,549,551,550,553],"class_list":["post-1872841","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-technologie","tag-ai-crawlers","tag-ai-roboti","tag-ai-search","tag-ai-vyhledavani","tag-anthropic","tag-anthropic-ai","tag-apple","tag-applebot","tag-automatizovany-provoz","tag-bingbot","tag-bot-management","tag-chatgpt","tag-chatgpt-user","tag-claude","tag-copilot","tag-ddos-ochrana","tag-duckduckbot","tag-duckduckgo","tag-facebook-crawler","tag-gemini","tag-google","tag-google-extended","tag-googlebot","tag-gptbot","tag-kyberneticka-bezpecnost","tag-llm","tag-meta","tag-microsoft","tag-oai-adsbot","tag-oai-searchbot","tag-ochrana-webu","tag-openai","tag-perplexity-ai","tag-perplexity-user","tag-perplexitybot","tag-proof-of-work","tag-rate-limiting","tag-web-crawlers","tag-web-scraping"],"_links":{"self":[{"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/posts\/1872841","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/comments?post=1872841"}],"version-history":[{"count":2,"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/posts\/1872841\/revisions"}],"predecessor-version":[{"id":1872999,"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/posts\/1872841\/revisions\/1872999"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/media\/1872956"}],"wp:attachment":[{"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/media?parent=1872841"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/categories?post=1872841"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/blog.wedos.com\/cs\/wp-json\/wp\/v2\/tags?post=1872841"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}