Xbox has today released its sixth Transparency Report spanning the period between January 1st and December 31st of 2025, and it contains all sorts of details about how Microsoft has moderated content on the Xbox platform.
The bit we're most interested in is towards the end — it talks about "Sharing our Safety Data, and mentions that a whopping 14.8 billion pieces of Xbox content were moderated proactively in 2025, of which 2.5% was "identified as containing harmful or policy-violating content". That 2.5% figure equals 368 million pieces in all.
"To reduce the risk of harm and prevent our players from being exposed to inappropriate content, we proactively scan content to identify and remove unwanted content before it impacts players.
We apply these methods to scan text, usernames, images, and other user generated content to ensure it avoids harmful behavior, respects others' rights, and maintains privacy. For example, proactive moderation allows us to find and remove inauthentic accounts and other cases of abuse of our platform and services."
150 million pieces were identified as "abusing our Platform and Services", while 62 million contained Profanity and Vulgarity, 51 million contained Nudity and Pornography, 38 million contained Bullying and Harassment, 28 million contained Hateful Conduct, 14 million contained Other Regulated Goods and Services, 13 million contained Graphic Violence and Human Gore, and 7 million contained Exposure of Personal Information. It goes on from there, but these are the categories with the most "violated policy" incidents.
Just to make it clear, these are proactive moderation efforts, which means players supposedly never saw most of them — they were detected by "AI capabilities" combined with Xbox's "existing proactive tooling" to filter them out.
"The expansion of AI capabilities combined with our existing proactive tooling has allowed us to increase and improve our detection of harmful content before it even reaches players."
Beyond this, 39 million reports were received from Xbox players manually in 2025, of which 3.7 million led to "restrictions imposed on content and accounts". Most of these were for Spam (1.7 million), but also Profanity and Vulgarity (555k), Viruses (412k), Bullying and Harassment (371k), Nudity and Pornography (270k), Hateful Conduct (225k), Abuse of Platform and Services (141k) and a few other things.
Altogether, around 14 million pieces of content were processed manually by Xbox in 2025, with the rest of the 350+ million being auto-processed. Here's a bit more about how crucial Xbox thinks this is:
"At Xbox, we believe automation and the use of AI-enabled solutions, combined with human expertise, play crucial and complementary roles in effectively identifying, reporting, and preventing harms at scale, especially as these online harms become more technologically sophisticated. They not only prevent unwanted content from reaching players, they also reduce human exposure to sensitive content and help focus human moderation efforts on more nuanced and complex issues. We have a number of safeguards in place to monitor automations. This includes a robust quality assurance program, moderator reviews, a suite of player reporting tools, and an appeals process."
To round off this section, Microsoft mentions that 1,254 reports were sent to the National Center for Missing and Exploited Children in 2025 ("as required by US law"), and 10,051 referral messages were sent to players via the Crisis Text Line or an international equivalent.
And finally, there were 332K unique appeals against bans or other enforcements on Xbox in 2025, 26% of which were successful appeals. The other 74% of cases were deemed to be legit and the enforcements were upheld.
This is all just one small part of the Transparency Report, so we recommend reading it in full if you're interested — and go and check out Xbox Wire's article on the topic as well, as they've got a lot more details on staying safe with Xbox and specifically practicing safer decisions with Minecraft.