We Scanned 1.9M Agents. 99.2% Are Fake.

18 platforms audited. 17 sybil reports published. The AI agent economy is 3,000–5,000 humans running scripts. ClawForce tells you which agents are real.

What We Found — Feb 8, 2026
1.9M
agent identities scanned across 18 platforms
99.2%
are sybils — confirmed via public API data
~5,000
estimated real human operators behind it all
Live Demo
Demo Data
Live API
Real agents from our audit:
VERIFICATION SCORE
Sybil Probability
Trust Level
Uniqueness
18 Platforms Audited — Feb 8, 2026
Tier 1 — Confirmed Sybil Infestation
PlatformTypeKey FindingSybil %
MoltBook Social 88 bots in 4 seconds. 96,798 comments on one post. 99.2%
claw-jobs.com Gig market 96% gigs from 1 poster. Phantom completions. 98.7%
ClawNews News 411 agents, only 32 verified (7.8%). 92.2%
ugig.net Freelance 10 gigs in 6 min from 1 account. Zero completed. ~86%
LobChan Forum Automated posting (5s gaps). Tripcode collision. 84%
ClawTasks Bounties 14 agents in 8 seconds. 11.77h completion fingerprint. 72%
Shipyard Deploy 15 agents at exact 13-second intervals. 60-70%
MoltCities Social ~60% inactive registrations. ~60%
clawtrade.net Trading 5 agents in 80ms. Wash trading confirmed. 59%
MoltLaunch Token market 1 user registered 3x, created 3 tokens. No prevention. 50%
MoltSlack Coordination "Mythology Swarm" — 10 agents in 3 minutes. 41-50%
molthub.studio Social 7 seed accounts at same millisecond. 79.7% of karma. 35-50%
Tier 2 — Structural Vulnerability
PlatformTypeKey FindingRisk
clawn.ch Token launch 51 bots launched 376 tokens. 205 Twitter sybil ring. HIGH
RentAHuman Human hire 100 profiles, 0 bookings. Non-functional. HIGH
DevAIntArt AI gallery 72.9% of art has no attribution. 1 operator controls 93%+. MEDIUM
Clawdmint NFT launch All 4 collections by 1 operator (LobsterGraphix). MEDIUM
Tier 3 — Pre-Growth
PlatformTypeKey FindingAgents
Moltverr Freelance API locked. Bidding vulnerable to sybils.
Molt Road Marketplace 421 agents, 12 listings. 4.58M MOLTROAD treasury. 421
All data collected from public APIs. No authentication required. Fully reproducible.
MILLISECOND BATCHES
clawtrade: 5 agents in 80ms
MoltBook: 88 bots in 4 seconds
ClawTasks: 14 agents in 8 seconds
molthub: 7 at same millisecond
PHANTOM MARKETPLACES
claw-jobs: 6/8 "done" gigs have no worker
RentAHuman: 100 profiles, 0 bookings
ugig: 0 completed on leaderboard
Zero functional marketplaces found
CROSS-PLATFORM LINKS
eltociear: 4 platforms, same tripcode
Kahbot: duplicate reg, 10s apart
LobsterGraphix: all 4 NFT collections
Same operators across ecosystem
How It Works

Sybil Detection

Build a social graph from agent relationships. Extract PageRank, betweenness centrality, clustering coefficient. Feed into Isolation Forest anomaly detector. Output: sybil probability 0–1. Legit agents score ~0.28. Bot clusters score 0.90.

Behavioral Fingerprinting

Generate 768-dimensional embedding vectors from agent text using RoBERTa. Compare via cosine similarity. Similarity above 0.92 flags sockpuppets. One test caught 99.7% text similarity between controller and puppet.

Trust Scoring

Exponential decay: trust = initial × e-0.01 × days. Active agents refresh trust through participation. Dormant agents lose governance weight. Prevents the "register and abandon" attack.

API
POST /sybil/score
Analyze agent social graph for sybil probability
// Request
{
  "agents": [
    { "agent_id": "mira", "follower_count": 45, "following_count": 32, "votes_cast": 18 }
  ],
  "edges": [
    { "source": "mira", "target": "sam_moltman", "edge_type": "follows" }
  ]
}

// Response
{
  "scores": [
    { "agent_id": "mira", "sybil_probability": 0.08, "risk_level": "low", "cluster_id": -1 }
  ],
  "model_version": "graph-iforest-v2"
}
POST /fingerprint/analyze
Generate behavioral fingerprint from agent text output
// Request
{
  "agent_id": "mira",
  "text_samples": [
    { "text": "Governance requires transparency...", "source": "proposal" }
  ]
}

// Response
{
  "agent_id": "mira",
  "uniqueness_score": 0.82,
  "similar_agents": [],
  "features": { "vocab_richness": 0.73, "avg_word_length": 5.2 }
}
POST /trust/decay
Calculate trust score with exponential decay and activity renewal
POST /voting/anomaly
Detect voting blocs and anomalous patterns via Shannon entropy
Quick start
curl -X POST http://localhost:8000/sybil/score \
  -H "Content-Type: application/json" \
  -d '{
    "agents": [{"agent_id": "test", "follower_count": 5, "following_count": 3}],
    "edges": []
  }'
Deployed On-Chain
124 passing smart contract tests
ContractAddressTests
MoltEscrow 0x8BB8...a8D4 24 passing
BuzzbotRegistry 0x72Eb...5c0 29 passing
NewsAttestation 0x29BB...32f9 24 passing
MedCredential 0xEEf8...F04B 21 passing
JobEscrow 0x858B...e5b1 26 passing
All contracts verifiable on BSC Testnet Explorer