Fair Play and Safety Glossary · Gaming

Aimbot, triggerbot: a cheat that aims or fires for the player using enemy positions and synthetic input; it sends valid inputs, so server authority can't stop it and detection has to (Anti-Cheat).
Anti-tamper: a protection layer, such as Denuvo, that makes the game binary and its DRM hard to reverse-engineer or modify; often used for a launch window and removed later (Anti-Cheat).
Auto-mute: blocking a zero-tolerance chat message before it's sent and muting the sender, as League's 2023 system does; filtering before delivery fails closed (Player Behavior & Moderation).
Botting: software simulating game input, usually unattended, to level accounts for sale or farm currency; Vanguard took League from over a million bot hours a day to under 5,000 (Anti-Cheat).
Cheat case: signals per account accumulated into one case with a score and a model version, so automated action happens only at high confidence and the middle band goes to human review (Anti-Cheat).
Children's risk assessment (OSA): the UK Online Safety Act assessment of risks to children that services likely to be accessed by them had to complete by July 24, 2025 (Player Behavior & Moderation).
Client integrity: client-side protection that makes the game hard to read or modify (encrypted code, randomized anti-debug checks that react late, encrypted values, an obfuscated protocol); it raises cheat costs and decays as cheats adapt (Anti-Cheat).
Content maturity label: Roblox's labels (Minimal, Mild, Moderate, Content May Vary, Restricted) that decide which age groups can play an experience; unlabeled experiences are restricted for everyone (Player Behavior & Moderation).
Delayed enforcement (ban wave): collecting signature detections and enforcing later, so cheat sellers learn which check fired as late as possible, while immediate harm in a ranked match is stopped at once (Anti-Cheat).
Disruptive behavior, harmful conduct: conduct that doesn't match the norms a player and community set, and the subset that causes significant emotional, mental or physical harm, per the Disruption and Harms in Online Gaming Framework (Player Behavior & Moderation).
DMA (direct memory access): reading a game's memory from a second device over PCIe, outside the operating system, which is why kernel anti-cheat can't see everything; not the EU's Digital Markets Act (Anti-Cheat).
DRM: technology that checks a copy of a game was paid for; it protects sales, while anti-cheat protects the fairness of matches (Anti-Cheat).
DSA Transparency Database: the European Commission's public database to which online platforms submit their statements of reasons (Player Behavior & Moderation).
Easy Anti-Cheat, BattlEye, VAC: Epic's prevention-first anti-cheat with free and paid tiers, a kernel-based protection system with backend-controlled scanning, and Valve's signature-based system that never bans on reports alone (Anti-Cheat).
Hardware ban: a ban on a hard-to-change device identifier, such as the TPM, so a banned player can't just make a new account; stored hashed, with an end date and an appeal path for secondhand and shared machines (Anti-Cheat).
In-game support SDK: a help center and ticket or messaging inside the game, filled with context the game already knows; on consoles it can hand off to a phone by QR code (Player Support).
In-match report: a report filed when the behavior happens, naming the match, channel and category, with the server attaching evidence from its own records and weighing the reporter's credibility (Player Behavior & Moderation).
Intentional feeding, AFK: losing on purpose by walking into the enemy, and going idle or leaving; detected from match telemetry by the game team, with frequency-based penalties for leaving (Player Behavior & Moderation).
Kernel anti-cheat: an anti-cheat driver running with the operating system's privilege, often loaded at boot, so user-mode cheats can't outrank it; it risks stability and abuse, and Proton doesn't support it on Linux and Steam Deck (Anti-Cheat; its risks are in The Risks of Kernel Drivers).
Notice and action: the DSA Article 16 mechanism that lets anyone flag illegal content to a hosting service, which must process the notice and tell the notifier the outcome; it covers game chat and UGC as well as IP claims (Player Behavior & Moderation; UGC & Creator Economies).
Online platform (DSA): a hosting service that disseminates user content to the public, which brings duties beyond notice and action such as internal complaint handling and submissions to the Transparency Database (Player Behavior & Moderation).
Operation (support): a named, idempotent action support takes on an account (grant, restore, reverse a takeover), capped by value and approved by a second person above a threshold; there is no operation for changing sign-in details or ending a ban (Player Support).
Ownership verification: a server-side check through the store's API that an account owns the game or DLC, the server-trusted form of a license check (Anti-Cheat).
Penalty ladder: escalating responses per behavior that restrict the abused feature first (chat, queue, ranked) before the account, with zero-tolerance violations skipping straight to a permanent ban (Player Behavior & Moderation).
Prevalence, time to action: the share of matches with a cheater by mode and region, and the matches from first detection to enforcement; Riot reported League dropping from 45+ games to under 10 after Vanguard (Anti-Cheat).
Prevention, detection, deterrence: the three parts publishers organize anti-cheat into (Riot names them; Valve's Steamworks guidance covers the same ground), with prevention preferred because it stops cheating before players meet the cheater (Anti-Cheat).
Proactive detection: scanning content before anyone reports it; Xbox scanned 14.8 billion items in 2025 and found far more violations than players reported (Player Behavior & Moderation).
Reform: the goal of most penalties, since most disruptive players are only sometimes disruptive; consequences work when they're specific, prompt and proportionate, with strikes that expire (Player Behavior & Moderation).
Request a Human: a button that hands a bot-answered support ticket to a person, which brought Riot's bot satisfaction up to match agents; denials stay with people (Player Support).
Scripting: third-party programs that attach to the game client to read its state and automate precise actions; as many as 1 in 15 League games had a scripter or botter before Vanguard (Anti-Cheat).
Secure Boot and TPM requirements: PC shooters requiring the platform's verified boot chain (Battlefield 6) or TPM 2.0 and Secure Boot (Call of Duty: Black Ops 7) so a cheat driver can't load before the anti-cheat (Anti-Cheat).
Shadow mode: running a moderation vendor against a labeled sample of your own reports, without acting, before it's allowed to decide anything (Player Behavior & Moderation).
Statement of reasons: the explanation DSA Article 17 requires a hosting service to give a user it restricts: the restriction, the facts, whether automation was used, the ground and the redress (Player Behavior & Moderation).
Trust and safety: the policy, intake, decisions, penalties, appeals, audit and reporting a product runs against abuse; the general system is the playbook's, and games add matches, real-time channels and gameplay detection (Player Behavior & Moderation; Trust & Safety).
Vanguard: Riot's anti-cheat, a user-mode client plus a kernel driver loaded at boot that sends nothing back, using TPM 2.0 as a hard-to-fake hardware identifier (Anti-Cheat).
Voice evaluation: moderating voice by recording on a short fuse and evaluating on report (VALORANT keeps recordings 24 hours, 7 days if reported) or by analyzing live channels and surfacing events with clips (ToxMod) (Player Behavior & Moderation).
Wallhack, ESP, map hack: cheats that show enemies, items or state the player shouldn't see, read from client memory; defeated by withholding hidden state (Anti-Cheat).
Withholding hidden state (Fog of War): the server not sending enemy positions until a client needs to display them, the general form being per-client relevancy as in Unreal; VALORANT does it with a precomputed potentially visible set at under 2% of server frame time (Anti-Cheat).
Zero tolerance: the categories (threats, sexual content involving minors, extremism) that skip the penalty ladder and, in chat, are auto-muted before they're sent (Player Behavior & Moderation; the ladder in Penalties and Reform).

Aimbot, triggerbot: a cheat that aims or fires for the player using enemy positions and synthetic input; it sends valid inputs, so server authority can't stop it and detection has to (Anti-Cheat).
Anti-tamper: a protection layer, such as Denuvo, that makes the game binary and its DRM hard to reverse-engineer or modify; often used for a launch window and removed later (Anti-Cheat).
Auto-mute: blocking a zero-tolerance chat message before it's sent and muting the sender, as League's 2023 system does; filtering before delivery fails closed (Player Behavior & Moderation).
Botting: software simulating game input, usually unattended, to level accounts for sale or farm currency; Vanguard took League from over a million bot hours a day to under 5,000 (Anti-Cheat).
Cheat case: signals per account accumulated into one case with a score and a model version, so automated action happens only at high confidence and the middle band goes to human review (Anti-Cheat).
Children's risk assessment (OSA): the UK Online Safety Act assessment of risks to children that services likely to be accessed by them had to complete by July 24, 2025 (Player Behavior & Moderation).
Client integrity: client-side protection that makes the game hard to read or modify (encrypted code, randomized anti-debug checks that react late, encrypted values, an obfuscated protocol); it raises cheat costs and decays as cheats adapt (Anti-Cheat).
Content maturity label: Roblox's labels (Minimal, Mild, Moderate, Content May Vary, Restricted) that decide which age groups can play an experience; unlabeled experiences are restricted for everyone (Player Behavior & Moderation).
Delayed enforcement (ban wave): collecting signature detections and enforcing later, so cheat sellers learn which check fired as late as possible, while immediate harm in a ranked match is stopped at once (Anti-Cheat).
Disruptive behavior, harmful conduct: conduct that doesn't match the norms a player and community set, and the subset that causes significant emotional, mental or physical harm, per the Disruption and Harms in Online Gaming Framework (Player Behavior & Moderation).
DMA (direct memory access): reading a game's memory from a second device over PCIe, outside the operating system, which is why kernel anti-cheat can't see everything; not the EU's Digital Markets Act (Anti-Cheat).
DRM: technology that checks a copy of a game was paid for; it protects sales, while anti-cheat protects the fairness of matches (Anti-Cheat).
DSA Transparency Database: the European Commission's public database to which online platforms submit their statements of reasons (Player Behavior & Moderation).
Easy Anti-Cheat, BattlEye, VAC: Epic's prevention-first anti-cheat with free and paid tiers, a kernel-based protection system with backend-controlled scanning, and Valve's signature-based system that never bans on reports alone (Anti-Cheat).
Hardware ban: a ban on a hard-to-change device identifier, such as the TPM, so a banned player can't just make a new account; stored hashed, with an end date and an appeal path for secondhand and shared machines (Anti-Cheat).
In-game support SDK: a help center and ticket or messaging inside the game, filled with context the game already knows; on consoles it can hand off to a phone by QR code (Player Support).
In-match report: a report filed when the behavior happens, naming the match, channel and category, with the server attaching evidence from its own records and weighing the reporter's credibility (Player Behavior & Moderation).
Intentional feeding, AFK: losing on purpose by walking into the enemy, and going idle or leaving; detected from match telemetry by the game team, with frequency-based penalties for leaving (Player Behavior & Moderation).
Kernel anti-cheat: an anti-cheat driver running with the operating system's privilege, often loaded at boot, so user-mode cheats can't outrank it; it risks stability and abuse, and Proton doesn't support it on Linux and Steam Deck (Anti-Cheat; its risks are in The Risks of Kernel Drivers).
Notice and action: the DSA Article 16 mechanism that lets anyone flag illegal content to a hosting service, which must process the notice and tell the notifier the outcome; it covers game chat and UGC as well as IP claims (Player Behavior & Moderation; UGC & Creator Economies).
Online platform (DSA): a hosting service that disseminates user content to the public, which brings duties beyond notice and action such as internal complaint handling and submissions to the Transparency Database (Player Behavior & Moderation).
Operation (support): a named, idempotent action support takes on an account (grant, restore, reverse a takeover), capped by value and approved by a second person above a threshold; there is no operation for changing sign-in details or ending a ban (Player Support).
Ownership verification: a server-side check through the store's API that an account owns the game or DLC, the server-trusted form of a license check (Anti-Cheat).
Penalty ladder: escalating responses per behavior that restrict the abused feature first (chat, queue, ranked) before the account, with zero-tolerance violations skipping straight to a permanent ban (Player Behavior & Moderation).
Prevalence, time to action: the share of matches with a cheater by mode and region, and the matches from first detection to enforcement; Riot reported League dropping from 45+ games to under 10 after Vanguard (Anti-Cheat).
Prevention, detection, deterrence: the three parts publishers organize anti-cheat into (Riot names them; Valve's Steamworks guidance covers the same ground), with prevention preferred because it stops cheating before players meet the cheater (Anti-Cheat).
Proactive detection: scanning content before anyone reports it; Xbox scanned 14.8 billion items in 2025 and found far more violations than players reported (Player Behavior & Moderation).
Reform: the goal of most penalties, since most disruptive players are only sometimes disruptive; consequences work when they're specific, prompt and proportionate, with strikes that expire (Player Behavior & Moderation).
Request a Human: a button that hands a bot-answered support ticket to a person, which brought Riot's bot satisfaction up to match agents; denials stay with people (Player Support).
Scripting: third-party programs that attach to the game client to read its state and automate precise actions; as many as 1 in 15 League games had a scripter or botter before Vanguard (Anti-Cheat).
Secure Boot and TPM requirements: PC shooters requiring the platform's verified boot chain (Battlefield 6) or TPM 2.0 and Secure Boot (Call of Duty: Black Ops 7) so a cheat driver can't load before the anti-cheat (Anti-Cheat).
Shadow mode: running a moderation vendor against a labeled sample of your own reports, without acting, before it's allowed to decide anything (Player Behavior & Moderation).
Statement of reasons: the explanation DSA Article 17 requires a hosting service to give a user it restricts: the restriction, the facts, whether automation was used, the ground and the redress (Player Behavior & Moderation).
Trust and safety: the policy, intake, decisions, penalties, appeals, audit and reporting a product runs against abuse; the general system is the playbook's, and games add matches, real-time channels and gameplay detection (Player Behavior & Moderation; Trust & Safety).
Vanguard: Riot's anti-cheat, a user-mode client plus a kernel driver loaded at boot that sends nothing back, using TPM 2.0 as a hard-to-fake hardware identifier (Anti-Cheat).
Voice evaluation: moderating voice by recording on a short fuse and evaluating on report (VALORANT keeps recordings 24 hours, 7 days if reported) or by analyzing live channels and surfacing events with clips (ToxMod) (Player Behavior & Moderation).
Wallhack, ESP, map hack: cheats that show enemies, items or state the player shouldn't see, read from client memory; defeated by withholding hidden state (Anti-Cheat).
Withholding hidden state (Fog of War): the server not sending enemy positions until a client needs to display them, the general form being per-client relevancy as in Unreal; VALORANT does it with a precomputed potentially visible set at under 2% of server frame time (Anti-Cheat).
Zero tolerance: the categories (threats, sexual content involving minors, extremism) that skip the penalty ladder and, in chat, are auto-muted before they're sent (Player Behavior & Moderation; the ladder in Penalties and Reform).