Why is an old image still shown in Kakao Talk?

Kakao caches scrap results. After checking the page's Open Graph tags, use the OG Cache Initialization Tool to remove the stored preview.

Does KakaoTalk Scrap respect robots.txt?

The bot metadata does not establish compliance, and Cloudflare's bot directory marks the bot as not following robots.txt. Treat the file as a preference unless logs confirm otherwise.

Does a scrap request represent a visitor?

No. It is a server request used to construct a Kakao Talk message and may occur without anyone opening the resulting link.

Is KakaoTalk Scrap used for AI training?

No training or AI search role is documented. Its recorded purpose is webpage scraping for Kakao Talk previews.

KakaoTalk Scrap & AI Search: What to Know

What is KakaoTalk Scrap?

KakaoTalk Scrap turns a webpage URL into a Kakao Talk scrap message. When a user or an application asks Kakao to scrape the address, Kakao's server retrieves the page and builds a feed-style message from its public web data.

Kakao's message-template documentation says the server reads Open Graph tags from the page head. og:image supplies the representative image, while og:title, og:description, and og:site_name fill the corresponding message fields. Kakao can cache that result, so a changed image or title may not appear until its OG cache is reset.

The request is part of composing or sending a Kakao Talk message. It is not a human pageview, and this bot gives the bot no AI search or model-training purpose. Allowing a scrape can affect the Kakao preview only; it does not provide a documented route into AI answers.

The full recorded user agent is kakaotalk-scrap/1.0; +https://devtalk.kakao.com/t/scrap/33984, with kakaotalk-scrap as the stable token. Robots compliance is not confirmed in the bot metadata, while Cloudflare's bot directory marks it false. No verified IP or signature-agent URL is attached to this identity.

Indirectly relevant

Is KakaoTalk Scrap relevant for AI search?

Indirectly. KakaoTalk Scrap has no AI product of its own, but its output can end up in the systems that AI answers draw on.

KakaoTalk Scrap fetches pages to build link-preview cards. It does not feed an AI index, but it reads the same Open Graph and structured metadata that AI systems parse to understand your pages, so clean preview metadata doubles as AI-friendly metadata.

How to handle KakaoTalk Scrap

Allow the scrap server to read public pages and image URLs that should appear in Kakao Talk messages. Keep internal pages behind authentication, and do not expose an image URL that is intended only for a private network.

The stable token supports this robots.txt preference:

User-agent: kakaotalk-scrap
Disallow: /

Cloudflare says the bot does not follow robots.txt, so verify the behavior or enforce a denial outside the file. For an allowed page whose preview is stale, correct its Open Graph values first and then use Kakao's OG Cache Initialization Tool.

Examples

A shop sends a product URL through a Kakao Talk scrap message, and Kakao maps the page's `og:image` and title into the message.
A publisher replaces an article image, then clears Kakao's OG cache after the old image continues to appear in newly created previews.

Frequently asked questions about KakaoTalk Scrap

Learn about AI visibility monitoring and how Promptwatch helps your brand succeed in AI search.

Kakao documents Open Graph values for the representative image, page title, description, and site name. It also retains the requested URL for the message link.