01
Player risk inference
Classify wallets and player sessions by fraud probability, bot behavior, exploit proximity, and account history.
Bitfragments AI Model
Security intelligence for autonomous game economies.
A risk model for Web3 games that detects bots, Sybil farms, suspicious wallets, exploit patterns, marketplace fraud, and reward abuse before they damage your economy.
24/7
live wallet and economy monitoring
<120ms
risk response target for gameplay flows
50+
security signals across game activity
BOT DETECTION / SYBIL ANALYSIS / WALLET RISK / REWARD GUARD / NFT FRAUD / MARKETPLACE WATCH / EXPLOIT MONITORING / BOT DETECTION / SYBIL ANALYSIS / WALLET RISK / REWARD GUARD / NFT FRAUD / MARKETPLACE WATCH / EXPLOIT MONITORING /
One model for game security decisions.
Plug the model into login, matchmaking, loot drops, marketplace listings, DAO voting, treasury actions, and reward claims to make abuse-aware decisions in real time.
02
Economy anomaly detection
Detect abnormal reward velocity, farming loops, whale manipulation, fake trading, and token inflation patterns.
03
Action policy engine
Return clear decisions: allow, challenge, throttle, flag, freeze, or escalate to your security team.
Signals built for GameFi.
The model fuses onchain transactions, gameplay telemetry, wallet graphs, marketplace behavior, NFT ownership, and session integrity into one decision layer.
From raw events to security actions.
Every claim, trade, login, vote, mint, bridge event, and reward action can be scored before the transaction finalizes or before an item is released.
1
Collect telemetry
Wallet, session, contract, marketplace, reward, inventory, and device signals.
2
Score behavior
AI model detects abuse clusters, suspicious timing, duplication attempts, and Sybil patterns.
3
Trigger policy
Your game receives a decision and explanation that can be enforced instantly.
risk-check.ts
// Bitfragments AI Model API
import { BitfragmentsAI } from "@bitfragments/ai";
const model = new BitfragmentsAI({
apiKey: process.env.BITFRAGMENTS_KEY,
model: "bf-ai-risk-v1",
});
const decision = await model.scoreAction({
player: "0xA18...9f2",
action: "loot_claim",
sessionId: "sess_8c91",
contract: "0xGameItems",
chainId: 8453,
signals: {
rewardVelocity: 0.74,
walletAgeDays: 3,
duplicateClaimDistance: 0.91,
marketplaceExposure: 0.22,
},
});
if (decision.policy === "challenge") {
return game.requestSessionProof(player);
}
if (decision.policy === "block") {
return security.freezeReward(decision.reason);
}
return game.releaseLoot(player);Where the model plugs in.
Use the AI model anywhere your game needs a fast security decision without ruining gameplay.
Block high-risk wallets, require extra verification, or isolate suspicious accounts.
Stop automated farms, duplicate claims, scripted drops, and abnormal earning loops.
Score suspicious transfers, fake ownership patterns, stolen assets, and wash activity.
Detect fake sales, price manipulation, royalty bypass, and coordinated marketplace abuse.
Identify Sybil voting, hostile takeovers, vote buying, and treasury attack preparation.
Escalate abnormal withdrawals, bridge events, contract interactions, and drain behavior.
Group coordinated players by behavior, timing, wallet graph, inventory, and sessions.
Stress-test token rewards, item sinks, inflation, and abuse incentives before release.