Sites blocking AI bots in Chile in 2026: only 13 of 92 close the door

Who is shutting the door on ChatGPT in Chile? Very few. On November 11, 2026 we checked the robots.txt file of 107 Chilean websites (media, banks, retail, telecom, health, universities and government) and of 84 media sites in six other countries. In Chile, only 13 of 92 sites with data block any AI bot, and just 10 block GPTBot, the crawler OpenAI uses to collect data for training its models. Meanwhile, 21 sites already publish an llms.txt file, the proposal to guide AI through a website. In Chile, more companies are opening the door to AI than closing it.

Robots.txt is a public text file at the root of every website that tells bots what they may crawl. Its format is standardized in RFC 9309. Blocking a bot there is a business decision: it keeps content out of model training or, depending on the bot, out of the answers of an AI search engine.

13of 92 Chilean sites block at least one AI bot

10block GPTBot, OpenAI's training bot

2block OAI-SearchBot, the ChatGPT search bot

21of 107 sites publish an llms.txt file

Source: ESBUENISIMO LINKS study: robots.txt and llms.txt of 191 sites checked on November 11, 2026

View the data as a table

of 92 Chilean sites block at least one AI bot	13
block GPTBot, OpenAI's training bot	10
block OAI-SearchBot, the ChatGPT search bot	2
of 107 sites publish an llms.txt file	21

Sites blocking AI bots in Chile: results by sector

Proportionally, no sector exceeds 20%: banks, telecom and health each reach 2 of 10 sites with data; media, 4 of 25 (16%); retail, 2 of 18; government, 1 of 8 (minsal.cl); and no university blocks anything. The difference is depth. The four media sites that block (El Mercurio's emol.com, Copesa's latercera.com and lacuarta.com, and publimetro.cl) shut out between 7 and 14 bots each. In other sectors, except Movistar, the Health Ministry and Mercado Libre, blocking is limited to 1, 2 or 3 bots.

The “with data” column excludes 15 sites whose robots.txt could not be read because the server blocked the request or returned a web page instead of the file. Eight large sites, such as scotiabank.cl, sii.cl and copec.cl, have no robots.txt at all: for bots, that means everything is open.

Banks and fintech20.0%

Telecom, energy and services20.0%

Pensions, health and insurance20.0%

Media16.0%

Government12.5%

Retail and ecommerce11.1%

Universities and institutes0.0%

Only sites with a readable robots.txt or none at all; sites that did not respond are excluded.Source: ESBUENISIMO LINKS study: robots.txt and llms.txt of 191 sites checked on November 11, 2026

View the data as a table

Banks and fintech	20.0% · 2/10
Telecom, energy and services	20.0% · 2/10
Pensions, health and insurance	20.0% · 2/10
Media	16.0% · 4/25
Government	12.5% · 1/8
Retail and ecommerce	11.1% · 2/18
Universities and institutes	0.0% · 0/11

Which AI bots get blocked, and why the difference matters

Not all bots do the same job. OpenAI lists three in its official documentation: GPTBot collects content to train models, OAI-SearchBot builds the ChatGPT search index and ChatGPT-User visits a page when a user asks for it. Google does something similar with Google-Extended: blocking it keeps content out of Gemini but does not affect Google Search.

GPTBot10

CCBot10

Google-Extended8

Bytespider8

ClaudeBot7

anthropic-ai5

Applebot-Extended5

meta-externalagent5

ChatGPT-User4

PerplexityBot4

OAI-SearchBot2

Source: ESBUENISIMO LINKS study: robots.txt and llms.txt of 191 sites checked on November 11, 2026

View the data as a table

GPTBot	10
CCBot	10
Google-Extended	8
Bytespider	8
ClaudeBot	7
anthropic-ai	5
Applebot-Extended	5
meta-externalagent	5
ChatGPT-User	4
PerplexityBot	4
OAI-SearchBot	2

The figure that matters most for visibility: of the 10 Chilean sites that block GPTBot, 8 still let OAI-SearchBot in. Emol, Mercado Libre, Movistar and the Health Ministry do not want their content to train models, but they do want to show up when someone searches in ChatGPT. Only La Tercera and La Cuarta also block the search bot. That distinction is worth checking on any site, because a misconfigured block can remove a brand from AI answers without anyone noticing. Our guide on configuring robots.txt for GPTBot and ClaudeBot covers each case.

Site	Sector	Bots blocked	GPTBot	llms.txt
lacuarta.com	Media	14	Blocked	No
latercera.com	Media	14	Blocked	No
minsal.cl	Government	9	Blocked	No
movistar.cl	Telecom, energy and services	9	Blocked	No
emol.com	Media	8	Blocked	No
mercadolibre.cl	Retail and ecommerce	7	Blocked	No
publimetro.cl	Media	7	Blocked	Yes
bancoripley.cl	Banks and fintech	3	Blocked	No
metrogas.cl	Telecom, energy and services	3	Blocked	No
clinicaalemana.cl	Pensions, health and insurance	2	Blocked	Yes
itau.cl	Banks and fintech	2	Allowed	Yes
easy.cl	Retail and ecommerce	1	Allowed	No
redsalud.cl	Pensions, health and insurance	1	Allowed	No
24horas.cl	Media	0	Allowed	No
abc.cl	Retail and ecommerce	0	Allowed	No
adnradio.cl	Media	0	Allowed	No
afpcapital.cl	Pensions, health and insurance	0	Allowed	Yes
afpmodelo.cl	Pensions, health and insurance	0	Allowed	No
aguasandinas.cl	Telecom, energy and services	0	Allowed	No
bancoestado.cl	Banks and fintech	0	Allowed	No
bancofalabella.cl	Banks and fintech	0	Allowed	Yes
bancointernacional.cl	Banks and fintech	0	Allowed	Yes
banmedica.cl	Pensions, health and insurance	0	Allowed	No
bcentral.cl	Government	0	Allowed	No
bci.cl	Banks and fintech	0	Allowed	No
bcn.cl	Government	0	Allowed	No
biobiochile.cl	Media	0	Allowed	No
cge.cl	Telecom, energy and services	0	Allowed	No
chileatiende.gob.cl	Government	0	Allowed	No
chilevision.cl	Media	0	Allowed	No
ciperchile.cl	Media	0	Allowed	No
clarochile.cl	Telecom, energy and services	0	Allowed	Yes
clinicalascondes.cl	Pensions, health and insurance	0	Allowed	No
cnnchile.com	Media	0	Allowed	No
colmena.cl	Pensions, health and insurance	0	Allowed	No
consalud.cl	Pensions, health and insurance	0	Allowed	No
cooperativa.cl	Media	0	Allowed	Yes
copec.cl	Telecom, energy and services	0	Allowed	No
cruzblanca.cl	Pensions, health and insurance	0	Allowed	Yes
cruzverde.cl	Retail and ecommerce	0	Allowed	No
df.cl	Media	0	Allowed	Yes
dt.gob.cl	Government	0	Allowed	No
duna.cl	Media	0	Allowed	No
duoc.cl	Universities and institutes	0	Allowed	No
elciudadano.com	Media	0	Allowed	No
eldesconcierto.cl	Media	0	Allowed	No
eldinamo.cl	Media	0	Allowed	No
elmostrador.cl	Media	0	Allowed	No
enel.cl	Telecom, energy and services	0	Allowed	No
entel.cl	Telecom, energy and services	0	Allowed	No
ex-ante.cl	Media	0	Allowed	Yes
falabella.com	Retail and ecommerce	0	Allowed	No
farmaciasahumada.cl	Retail and ecommerce	0	Allowed	No
fintual.cl	Banks and fintech	0	Allowed	Yes
gob.cl	Government	0	Allowed	No
hites.com	Retail and ecommerce	0	Allowed	No
inacap.cl	Universities and institutes	0	Allowed	No
jumbo.cl	Retail and ecommerce	0	Allowed	Yes
lanacion.cl	Media	0	Allowed	No
lapolar.cl	Retail and ecommerce	0	Allowed	No
lider.cl	Retail and ecommerce	0	Allowed	Yes
lun.com	Media	0	Allowed	No
mach.cl	Banks and fintech	0	Allowed	No
meganoticias.cl	Media	0	Allowed	Yes
paris.cl	Retail and ecommerce	0	Allowed	No
pcfactory.cl	Retail and ecommerce	0	Allowed	Yes
provida.cl	Pensions, health and insurance	0	Allowed	No
radioagricultura.cl	Media	0	Allowed	No
ripley.cl	Retail and ecommerce	0	Allowed	No
salcobrand.cl	Retail and ecommerce	0	Allowed	No
santaisabel.cl	Retail and ecommerce	0	Allowed	Yes
scotiabank.cl	Banks and fintech	0	Allowed	No
sernac.cl	Government	0	Allowed	No
sii.cl	Government	0	Allowed	No
sodimac.cl	Retail and ecommerce	0	Allowed	Yes
soychile.cl	Media	0	Allowed	No
t13.cl	Media	0	Allowed	No
tenpo.cl	Banks and fintech	0	Allowed	No
theclinic.cl	Media	0	Allowed	No
tottus.cl	Retail and ecommerce	0	Allowed	No
uai.cl	Universities and institutes	0	Allowed	Yes
uc.cl	Universities and institutes	0	Allowed	No
udd.cl	Universities and institutes	0	Allowed	Yes
udec.cl	Universities and institutes	0	Allowed	No
ufro.cl	Universities and institutes	0	Allowed	No
unab.cl	Universities and institutes	0	Allowed	Yes
unimarc.cl	Retail and ecommerce	0	Allowed	No
usach.cl	Universities and institutes	0	Allowed	No
usm.cl	Universities and institutes	0	Allowed	No
uv.cl	Universities and institutes	0	Allowed	No
vtr.com	Telecom, energy and services	0	Allowed	No
wom.cl	Telecom, energy and services	0	Allowed	No

“Bots blocked” counts 15 known AI bots. Excludes 15 sites whose robots.txt did not respond.Source: ESBUENISIMO LINKS study: robots.txt and llms.txt of 191 sites checked on November 11, 2026

Latin American media blocking AI bots: Chile is more open than its neighbors

For comparison, we checked 84 media sites in Mexico, Colombia, Peru, Argentina, Uruguay, Ecuador and Spain. The gap between countries is large, and there is a reason: blocking is decided by media group, not site by site.

0.0%10.0%25.0%40.0%55.0%–64.3%

Between 6 and 25 media sites per country. Countries without color were not measured. Spain (70%) was measured separately.Source: ESBUENISIMO LINKS study: robots.txt and llms.txt of 191 sites checked on November 11, 2026

View the data as a table

Colombia	64.3% · 9/14
Peru	50.0% · 6/12
Ecuador	50.0% · 4/8
Chile	16.0% · 4/25
Mexico	13.3% · 2/15
Argentina	0.0% · 0/15
Uruguay	0.0% · 0/6

In Peru, four Grupo El Comercio sites (gestion.pe, trome.com, ojo.pe and diariocorreo.pe) block the same nine bots, although elcomercio.pe, the group’s flagship paper, blocks none. In Colombia, 9 of 14 media sites block GPTBot, including eltiempo.com and portafolio.co. At the other end, no Argentine media site in the sample blocks GPTBot, while in Spain 7 of 10 do, several with exceptions for some sections. Chile sits in the lower middle: 4 of 25 media sites.

llms.txt: the Chilean sites talking to AI

llms.txt is a 2024 proposal: a Markdown file that summarizes what a site is and lists its key pages for language models. It is not an official standard and no major AI has confirmed it uses it for ranking, but adoption shows who is thinking about AI. In Chile, 21 of the 107 sites checked have one, including df.cl, cooperativa.cl, lider.cl, jumbo.cl, sodimac.cl, bancofalabella.cl, itau.cl, fintual.cl and Universidad del Desarrollo. One of them, publimetro.cl, publishes llms.txt while blocking GPTBot: it invites AI to read it, but not to train on its content.

What it means for a brand that wants to appear in AI

For companies that are not media outlets, the takeaway is simple: almost none are blocking AI, so the question is not whether bots can read the site but what they find there and what other sources say about the brand. In our study of the sources AI cites by industry in Chile, more than half of all citations went to sites other than the brands being evaluated.

For media, blocking has costs and benefits. Blocking training protects content; blocking search removes the outlet from ChatGPT answers, and with it the brands mentioned in its stories. A mention in a media site that allows OAI-SearchBot can reach a ChatGPT answer; one in a site that blocks it cannot, at least not through that route.

Three practical checks for any website:

• Open your own robots.txt and confirm that OAI-SearchBot, PerplexityBot and Googlebot are not blocked by mistake.
• Check whether bot protection on your server or CDN blocks AI crawlers even when robots.txt allows them. In this study, 15 Chilean sites did not return the file to a normal request.
• Measure whether the brand appears in AI answers, with a technical review of how AI crawlers read your site and mention tracking.

If you want to know what ChatGPT, Gemini and Perplexity say about your brand and which sources they use, the ESBUENISIMO AI Visibility platform measures it question by question, and our GEO agency works to get the brand into the sources AI actually reads.

Methodology

On November 11, 2026 we downloaded the robots.txt of 191 sites (107 Chilean sites and 84 media sites from other countries) using a standard browser user agent. A bot counts as blocked when a rule for that bot, or the general “*” rule when there is no specific one, disallows the site root (“Disallow: /”); in Spain, several sites block the root with exceptions for specific sections, and those count as blocks. Partial folder blocks are not counted. Sites without robots.txt (404 error) count as open, and sites that did not respond are excluded. llms.txt was counted when /llms.txt returned a text file rather than a web page. Robots.txt files can change at any time, and the list is not a statistical sample of all websites in each country.