Apex Legends developers, Respawn Entertainment have outlined their anti-cheat plans following an increase in cheating reports in the last few weeks. The devs said that they are using machine learning to detect and auto-ban cheaters.
In the latest Respawn Check-In on Reddit, Respawn's Community Manager Jay Frechette updated Apex Legends players on what exactly the developers have been working on in the past week.
There are a lot of changes and fixes ready to roll out in the next patch as you can see in the Apex Tracker on Trello, but the biggest talking point of the latest Check-In was an increased number of cheating reports, and what type of measures Respawn have in place to tackle these issues.
Frechette said that Respawn are aware of reports of cheaters, especially in Ranked mode, where cheating has been a hot topic in the previous weeks. Therefore, the team wanted to provide a little visibility into some of the work that's been going on behind the scenes to tackle cheating.
According to Frechette, Respawn is another development studio that use machine learning techniques to find cheaters. They create behaviour models that detect and even auto-ban players who use illegal tools to gain an unfair advantage over others.
They also worked on improving detection that identifies and bans new spam accounts even before they are used. Furthermore, the work on adapting to new cheats is always ongoing says Frechette.
Respawn have a matchmaking system that matches detected cheaters and spammers together, so they can use their hacks on each other. The devs are still investigating how people party up for matches. They say that even if you're not specifically using a cheat, partying up with cheaters is still considered cheating.
To make their fight against cheaters more effective, Respawn are increasing resources, whether that be people or tech. The war against cheaters continues to be an ongoing one for Respawn who made it a high priority in Apex Legends development.
You can find Frechette's latest update on Apex Legends subreddit.