Counter-Strike’s anti-cheat vocabulary has become genuinely hard to follow. A decade ago there was VAC and, later, Overwatch, and that was the entire conversation. Today a single CS2 forum thread can mention VAC, VAC Live, VACnet, Trusted Mode, Trust Factor, the Labeling Portal and ESIC within ten posts, all describing different mechanisms with overlapping names and completely different consequences.
Here is the whole stack in plain language, in the order a normal player will actually encounter it.
1. VAC: the slow permanent one
VAC, Valve Anti-Cheat, is the original system and still the one with the sharpest teeth. It detects cheat software and issues a ban that is permanent, cannot be removed by Steam Support, and applies to the game it was issued in plus titles sharing the same engine. Its defining characteristic is not detection but delay. VAC accumulates verdicts quietly and then bans tens of thousands of accounts at once, because if bans were instant, cheat developers could use their own accounts as a stopwatch and learn immediately whether a build had been caught.
2. VAC Live: the fast one
VAC Live arrived with CS2 and does what its name implies. When it is confident enough that someone in your ongoing match is cheating, it ends the game immediately rather than waiting for a wave. Honest players get their rating back; the suspect gets disconnected and dealt with later.
It also has the consequence players feel most viscerally: the cooldown. Valve’s documentation states that a player flagged for irregular gameplay receives a cooldown of at least one day across all modes, with their CS Rating adjusted — and that everyone in their party shares the punishment. If a match has ever vanished under you with a message about a VAC issue, this was the mechanism.
3. VACnet: the brain
VACnet is the machine-learning model Valve unveiled at GDC 2018, and it does the analytical work rather than the punishing. Instead of scanning your machine for suspicious processes, it examines gameplay data on the server and asks whether what this player is doing looks human. Its original training data came from Overwatch verdicts, which is why it still behaves with something resembling human caution.
4. Overwatch and the Labeling Portal: the humans
Overwatch was CS:GO’s community review system, where trusted players watched demos of suspects and voted. CS2 launched without it, brought back an invite-only version in the 25 April 2024 update, and then in August 2026 replaced the concept with something subtler: a browser-based VACnet Labeling Portal where invited reviewers classify clips into aim assistance, wall hacks, automated bunny hopping and bot behaviour, and where — importantly — the reviewer’s verdict does not directly ban anybody. The humans are generating training data now, not sentences.
5. Trust Factor: the bouncer
Trust Factor bans nobody and is arguably the most effective thing in the list. Introduced in CS:GO in November 2017, it scores your account on history and behaviour, then quietly decides who you get matched with. A good score means smooth lobbies and short queues; a poor one means a lobby full of suspicious accounts and a longer wait. It is invisible, unappealable, and it shapes every single match you play.
How they fit together
The easiest mental model is a funnel. Trust Factor sorts the population so that suspected bad actors mostly play each other. VACnet watches gameplay and identifies the suspicious. Overwatch and the Labeling Portal give humans a role in judging or labelling those cases. VAC Live can act during your match if the confidence is high enough. And VAC delivers the permanent verdict eventually, in a wave, weeks after the player annoyed you.
Nothing in this stack is perfect and several pieces are invisible by design. But knowing which layer did what makes CS2 far less maddening. Your match cancelled? That is VAC Live. Your cheater from three weeks ago finally vanished? That is VAC doing exactly what it was built to do. Your lobby full of suspicious accounts all night? That one is Trust Factor, and it is worth thinking about what your own account might be telling it.


Leave a Reply