Telegram prohibits collecting data for AI use

Telegram added an explicit prohibition against data scraping that applies to ‘all users, businesses, and third-party services accessing the platform’. The text refers to the Content Licensing and AI Scraping Terms which explicitly targets data acquisition to ‘train, fine-tune, validate or otherwise engage in the development, enhancement, benchmarking or deployment of artificial intelligence, machine learning models and similar technologies’.

Telegram is a messaging app founded by Pavel Durov and his brother Nikolai. It was marketed as a privacy-focused and secure app. However, this change also serves as a reminder that group chats are not end-to-end encrypted. For example, bot developers can access chat histories and ongoing conversations in order to provide their services. Chatbots can use user interactions to gather data and improve their services in a way that is indirectly related to the user (such as training data and fine-tuning). This may lead to the unintentional disclosure of sensitive information further along the AI pipeline. Despite security concerns and the availability of official alternatives, Telegram was notoriously popular with French officials.

A similar modification has also been detected in the developer terms.