Started by people tired of the same three gaps

Cerberus wasn't pitched as a moonshot. It started as a handful of security engineers comparing notes on where every anti-cheat product they'd worked with or been beaten by kept falling short.

Cerberus was founded in October 2024 by a small group of engineers with backgrounds in kernel-mode security products — EDR sensors, HIPS drivers, and independent hardware-cheat research. Between them they'd spent years watching the same pattern play out: a signature database catches a public cheat, a paid provider ships an undetected build within hours, and the anti-cheat vendor is back to reacting. Meanwhile the fixes on offer were either a heavier always-on kernel driver watching everything on the machine, or no hardware-layer defense at all against DMA boards and spoofed peripherals. Neither felt like an actual answer.

The founding idea was narrow on purpose: build detection across four independent layers — kernel integrity, behavioral AI, hardware fingerprinting, and network-side correlation — so that beating one layer doesn't mean beating Cerberus, and load the kernel component only while the protected game is running instead of leaving it resident all the time. The first year was almost entirely R&D: no customers, a handful of engineers, and a lot of internal red-teaming against our own driver before we trusted it near a live game process.

Closed beta opened in October 2025 with a small number of partner studios willing to run an early build in exchange for direct access to the team. We're still in that closed beta today, still adding studios one at a time rather than opening the gates all at once, and still fixing the things beta is supposed to surface. The team has grown from that original handful to 32 engineers across five departments, but the goal hasn't changed — close the three gaps above without pretending we've closed every gap that exists.

OCT 2024
Founded by a small kernel-security and threat-research team. Four-layer architecture drafted, R&D begins.
NOV 2024 – JAN 2025
Behavioral AI and SDK leads join. Internal red-teaming of the kernel driver, no external customers yet.
OCT 2025
Closed beta opens with the first partner studios across US-East and EU-West.
2026
AP-Southeast added, team grows to 32 engineers / 5 departments. Still closed beta, still v0.x.

What we try to hold to while we're still small

None of these are finished commitments — they're the standard we're building against, and the Trust Center shows exactly where we stand on each one today.

🛑

No telemetry by default

The Cerberus SDK isn't a background service. It loads with the protected game process and unloads when the game closes — nothing keeps collecting data once you've stopped playing.

📊

Honest about limitations

We publish our real uptime (97–98%, not 99.9%), our real detection rates (58–92% depending on category), and our real false-positive rate (0.14%) on the same pages a customer would see — not just the good months.

🔎

Security research first

The four-layer design came out of reverse-engineering actual DMA boards, firmware, and behavioral bypass tools — not a checklist. Threat research drives what we build next, not the other way around.

🐢

Slow, deliberate rollout

We add partner studios one at a time, our driver is EV code-signed with WHQL certification still pending, and our SOC 2 Type II audit is in progress, not complete. We'd rather say "not yet" than overstate where we are.

Where things stand today

Pulled from the same live engine that drives our Analytics and Status pages, not a static press-kit number.

32
Engineers
4
Detection Layers
3
Regions
Signatures Tracked

Want to see it running on your game?

We're still small, still early, and still adding partner studios one at a time. If the gaps above sound familiar, we'd like to talk.

Request Early Access