Fogo Foundation
Why it happened
Attack class: Access controlFoundation wallets validly signed 400M FOGO to a fresh wallet.
What would have stopped it
- Require independent multisig approval on every large treasury transfer.
- Allow-list treasury destinations and delay transfers to new addresses.
Not scored before the exploit
BlackHart had not scored Fogo Foundation before this exploit.
Someone took over Fogo Foundation wallets and sent 400 million FOGO, about $3 million and 4 percent of genesis supply, to their own wallet on August 28. The Foundation has not said how. User funds on the network were not affected, and the chain kept running until validators halted it the next day to freeze the attacker's addresses. When Fogo restarted on September 2 it had recovered and removed 237 million of the stolen tokens. The other 163 million, about $1.2 million, had already been spread across other wallets and is being chased with exchanges and law enforcement.
Two Fogo Foundation wallets sent 400 million FOGO to a fresh receiving wallet, the largest piece in a 240 million FOGO transfer, before about half was spread out through a dispersal wallet.
Attack flow
All 5 on-chain stepsThe attack in 4 steps. Step 1, Root control: Foundation wallet A (source), Foundation wallet B (source). Step 2, Collect: Receiving wallet (hub), about $3M reached this step. Step 3, Dispersal: Dispersal wallet (hub), about $1.5M reached this step. Step 4, Halt: Restricted and removed (exit), about $3M reached this step.
Step 1: Root control
- SourceFoundation wallet A160,000,000 FOGO~ $1.2MGbijjmVW2j...Mja8sw
- SourceFoundation wallet B240,000,000 FOGO~ $1.8M6h4Xw1eEDi...ps2t9i
Step 2: Collect
- HubReceiving wallet400,000,000 FOGO~ $3.0M6yxLzdRNQ9...77jcZH
Step 3: Dispersal
- HubDispersal walletabout 200M FOGOdozens of onward transfersAMbsex79gp...yr12mH
Step 4: Halt
- ExitRestricted and removed237M FOGO taken out of supply163M FOGO still outstanding
Why each step worked
- Foundation signing compromisedvector not disclosed
- No second approvermissing control400M FOGO to a wallet created that day
- No destination allow-list or delaymissing controlfresh address accepted at once
- Eleven hours to disclosuremissing controloutflows not caught in real time
- Validator halt and address restrictionsmall validator set upgraded the chain
Full forensic detail
Step-by-step reconstruction, root cause, counterfactuals, remediation, and disclosure timeline.
Exploit anatomy
Root cause
The Fogo Foundation lost control of who could authorize transfers from its treasury wallets. The Foundation says an unknown actor compromised the organization; it has not published the vector. No Fogo program or consensus flaw was involved: the chain was not attacked and kept running normally for about fifteen hours after the theft. On-chain, the 400 million FOGO left two Foundation wallets in transactions that each wallet validly signed, in the same pattern the Foundation used for its routine transfers (a small test, then the main amount, through durable-nonce transactions). The second wallet was topped up minutes before its 240 million FOGO transfer. Both wallets kept being used for Foundation transfers after the restart. That pattern fits an attacker steering the Foundation's own signing workflow, but it does not show whether keys, signing infrastructure or staff were the point of entry, and Fogo has not said. What contained the loss was not a wallet control but chain-level power: a small validator set halted the network and upgraded it to restrict the attacker's addresses, which let Fogo remove 237 million of the 400 million FOGO from supply.
Prevention analysis
A second approver checking the destination would have had to sign 400 million FOGO going to a wallet created that day.
The receiving wallet had never been used before. A delay on first-time destinations would have held the funds long enough to catch the theft before any FOGO moved on.
About 163 million FOGO left the receiving wallet in the eleven hours before disclosure. Faster detection would have shrunk the part that could not be recovered.
Similar incidents
Foundation-linked token wallets were drained in one burst, and the team voided the stolen tokens with a protocol-level freeze rather than a wallet control.
Privileged operational signing was turned against the protocol, with pre-signed durable-nonce transactions in the picture. Same class: weak control over what the organization's own signers approve.
Remediation
Timeline
Get your protocol scored across 12 dimensions, or request ongoing coverage.