Cerberus is still in closed beta — these are the first studios running it in production, and the honest, modest numbers they've seen so far. No inflated claims, just what actually happened in the first few weeks.
18 protected titles across 12 partner studios · 5 studios published case studies · v0.5.x beta channel · v0.6 alpha available
Partner Results
What changed after rollout
Every studio below opted into our closed beta knowing the product is early. Results vary by title, region, and how long a studio has been running Cerberus — we've asked each partner for numbers as of their most recent full month.
Ashfall Protocol's ranked ladder was being undermined by DMA cheat devices and Cronus-style hardware — the kind of cheating that usermode-only detection couldn't see, since the hardware sits outside the game process entirely. Vantage Point's internal reports of "suspiciously perfect" ranked matches had been climbing for two seasons.
Solution
Deployed Kernel Integrity (Layer 1) for syscall tamper detection and PCIe/DMA timing-anomaly analysis from Hardware Fingerprinting (Layer 3), paired with IOMMU/VT-d attestation on supported systems.
Results as of Aug 2026
Detected and banned a DMA-cheat ring of 11 linked accounts within 72 hours of Layer 1 rollout
Reported cheating incidents in ranked play down roughly 40% in the first month
No confirmed DMA bypass reports in the six weeks since — though the team is clear that's a beta-window result, not a permanence claim
"We knew DMA cheats existed in our game, we just couldn't see them. Cerberus caught a ring we'd been chasing for months within three days of going live. It's not a silver bullet, but it's the first anti-cheat that's actually shown us the hardware layer."
— Priya Malhotra, Lead Anti-Cheat Engineer, Vantage Point Games
A cluster of Discord servers and storefronts had grown up specifically around Riftbound Arena, selling aim and wallhack tools tuned to the game's client. Cindercrest's community team could see the sellers advertising in their own forums but had no reliable way to identify buyers before matches were already decided.
Solution
Rolled out Behavioral AI (Layer 2) — TCN-based aim analysis and trigger-timing spectral decomposition — alongside Network Sentinel (Layer 4) cross-player anomaly correlation to flag accounts tied to known seller signatures.
Results as of Aug 2026
18 accounts tied to a specific cheat-seller's build were identified and banned in the first two weeks
Ranked-queue cheating complaints in their support queue dropped from roughly 50/week to around 15/week
One tracked seller storefront stopped advertising builds "safe for Riftbound Arena" after the second ban wave
"The sellers were operating in the open, tagging our own subreddit in their ads. We're not naive enough to think we shut the ecosystem down — but we made it materially harder and slower for them, and our players noticed within days."
— Marcus Webb, Community & Trust Lead, Cindercrest Studios
Molten Anvil's previous anti-cheat vendor had a false-positive problem that spilled into public forums — legitimate players banned over unusual but innocent hardware or input patterns. By the time they evaluated Cerberus, community trust in any ban decision was low, and appeals volume was a full-time job for one support staffer.
Solution
Migrated fully to Cerberus's four-layer stack with an emphasis on the confidence-scoring pipeline — auto-ban only above a high confidence threshold, with mid-confidence cases routed to flagged secondary review instead of an immediate ban.
Results as of Aug 2026
False-positive appeals dropped from several per week under the old vendor to near zero over their first full month on Cerberus
Reinstated 6 accounts that had been wrongly banned by their previous anti-cheat, with zero new false-positive bans reported since migration
Support ticket volume tagged "unfair ban" fell noticeably, though Molten Anvil is watching this over a longer window before calling it settled
"Our old anti-cheat banned a top-100 leaderboard player for using an unusual controller. That's the kind of thing that costs you a community's trust for a year. We're six weeks into Cerberus and haven't had a single ban that made us wince yet."
— Sofia Andrade, Live Ops Director, Molten Anvil Interactive
Skybreak Royale's previous anti-cheat only had solid backend coverage in North America. Players in Southeast Asia and Western Europe were reporting cheaters going unpunished for days, and match-level detection was inconsistent enough that Paperlight's ops team suspected regional blind spots rather than a detection quality problem.
Solution
Adopted Cerberus across all three supported regions (US-East, EU-West, AP-Southeast) so Network Sentinel's packet-integrity and cross-player correlation checks run close to the match server instead of round-tripping to a single home region.
Results as of Aug 2026
AP-Southeast cheating reports fell an estimated 35% in the first month after regional coverage went live there
Time-to-ban for confirmed cheaters in EU-West and AP-Southeast is now roughly in line with their US-East numbers, closing a gap that had persisted for over a year
Regional uptime has held in the high 97% range across all three regions since launch — solid, not flawless, and Paperlight has said as much publicly
"We'd effectively been running a two-tier anti-cheat experience — good in NA, rough everywhere else. Getting real regional coverage in AP-Southeast and EU-West without renegotiating three separate vendor contracts was the actual win here."
— Devon Okafor, Technical Director, Paperlight Games
Undercroft Tactics · Tactical extraction shooter
BAN EVASION / REPEAT OFFENDERS
Halifax, Nova Scotia·Founded 2020·23 people·Website ↗
Integration: v0.5.x beta → v0.6 alpha (Sep 2026)
Challenge
Undercroft Tactics has a small, tight-knit competitive scene, which meant banned cheaters kept coming back on fresh accounts within days. Account-level bans alone weren't solving the actual problem — the same handful of people, cycling through new emails and hardware resets.
Solution
Added Hardware Fingerprinting (Layer 3) — hashed device and TPM 2.0 attestation signals used specifically for ban-evasion detection — on top of their existing behavioral detection, so repeat offenders get flagged even after creating a new account.
Results as of Aug 2026
Identified and banned a linked network of 40+ accounts tied to a single persistent ban-evasion group
Repeat-offender re-registration (same person banned again within 30 days) down an estimated 30% since rollout
Studio notes this is a cat-and-mouse process — a handful of evasion attempts still get through, and they expect that to continue as long as hardware spoofing tools evolve
"Banning the account was never the hard part. Stopping the same person from being back in ten minutes with a new email was. We're not going to pretend it's solved forever, but the group that used to live in our lobbies has mostly gone quiet."
— Elena Kowalski, Studio Head, Driftwood Signal
Thinking about closed beta for your studio?
We're still small, still early, and still adding partner studios one at a time. If any of these problems sound familiar, we'd like to talk.