Facebook asks its moderators to flag “locally illegal” posts in India, looking for 20 markers such as content on Kashmir, deities, and the Indian flag, The Indian Express reported on Monday, citing internal documents.

This contradicts the social media site’s publicly stated global policy, which does not consider speech attacking a religion or belief as hate speech, even in India, the daily said. The company also insists it does not block “locally illegal” content unless it finds a local government request “valid”.

Facebook did not respond to the newspaper’s queries on India-specific guidelines and if more countries have similar local guidelines.

The internal documents showed that Facebook’s content reviewers in India use the 20 indicators to mark posts for further review without the users’ knowledge or without any input from law enforcement agencies.

“Locally illegal” posts have content that violates Facebook policy or local laws “when the government actively pursues enforcement”, the documents showed. Content that risks getting Facebook blocked in a country is also defined as “locally illegal”.

Under the “national border” category for such content, reviewers are asked to flag posts that are “supportive” of a separate Kashmir state, on Pakistan’s claim to Kashmir and Siachen, China’s claim to Aksai Chin, posts on Arunachal Pradesh, Assam, Nagaland, or Tripura. “Azad Kashmir, Free Kashmir, Kashmir belongs to Pakistan… Look for maps invading territories, people protesting, etc,” the document says.

Under the “religious extremism” section, there are at least three instances where the phrase “humour is not allowed” is repeated.

The “defamation of deities” section in the religion category flags “negative remarks” or content “mocking images about religious gods and prophets”, “comparing deities” and “calling for new states based on religious community”.

“Legal is more risk-averse when it comes to litigation over religious imagery... If there is a violating image but the caption or context clearly condemns this defamation, such an image may still be considered offensive and liable to be GEO IP Blocked,” the document said.

Imagery of burning, stamping, and writing on the flag or depicting only a portion of the flag are also to be marked for review, The Indian Express reported.

The document on content moderation comes after the Centre’s draft rules make it mandatory for online platforms to help the government trace the origin of “unlawful content” whenever requested by authorities.

The social media group’s transparency reports showed that requests by Indian agencies to take down content has reduced over time. India had requested the platform to take down around 30,000 pieces of content in 2015, which came down to 3,000 in 2017. The report stated that most of the content that the government wanted to be taken down were “anti-religion” and “anti-state” posts.