Xbox has released its fourth Transparency Report, detailing the advancements the company’s gaming arm continues to promote player safety while addressing online toxic behaviour. The fourth Xbox Transparency Report highlights the company’s further use of AI throughout July to December 2023.
As reported in the latest Xbox Wire post, the company is continuing to invest in the use of AI in regards to combat toxicity amongst its online communities. It’s explained that Microsoft’s Responsible AI Standard is guiding the use of AI. According to Xbox, the use of AI is being used to assist human moderators to “focus on the more complex and nuanced harms.”
The company goes on to detail that early AI investments include ‘Auto Labeling’ and ‘Image Pattern Matching.’ Regarding the former, Xbox is leveraging AI to “classify conversational text by identifying words or phrases that match criteria and characteristics of potentially harmful content.” It’s explained that AI is able to analyze said text and assist community moderators on whether they can be categorized as harmful or inappropriate or if they are falsely reported. As far as Image Pattern Matching, AI is able to remove harmful content and “emergent toxic imagery rapidly” thanks to an advanced database and image-matching techniques the software can lean on.
Last year, Xbox released its third Transparency Report in which the use of AI was initially brought up. At the time, Xbox explained that AI would be used to organize and filter billions of interactions online. Xbox also began using the Turing Bletchley v3 multi-lingual model to scan user-generated imagery to ensure it was appropriate.
Outside of its advancements in AI, Xbox’s latest Transparency Reports state voice chat has “meaningfully improved” since the introduction of its voice reporting feature. It’s said 138,000 voice records have been captured using the company’s ‘Capture Now, Report Later’ system. Looking at the Enforcement Strike policy, 98% of players reportedly improved their behaviour, resulting in no subsequent enforcement. Apparently, 88% of Enforcement Strike offenders did not break Community Standards to result in another strike.
Finally, when it comes to blocking inauthentic accounts and malicious accounts, Xbox reports the number of proactive enforcements on inauthentic accounts have dropped for the first time in two years. Throughout the second half of 2023, only 7.3 million enforcements were made, down from the previous 16.3 from the prior report.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
