Christopher Grilhault des Fontaines retweeted
🇯🇵 In Japan, trust is built over decades.
NEC has spent more than 125 years building it with financial institutions.
Today, @NEC_jp and @dfnsHQ are partnering to explore digital asset infrastructure for Japanese financial institutions 🤝
The two companies have signed a ding to explore how DFNS can fit into NEC’s systems, including its planned custody offering.
The goal is to help Japanese institutions adopt digital assets without compromising their security, compliance, data residency or existing systems.
The partnership will focus on three areas:
1️⃣ Integrating DFNS wallets, key management, policies and transaction management into NEC’s digital asset infrastructure.
2️⃣ Adapting the platform to Japanese regulations and each institution’s security and operating requirements.
3️⃣ Supporting different deployment models, from SaaS to hybrid and fully on-premises, including air-gapped environments.
Institutions will be able to keep their keys, data and operations in Japan, in line with their requirements.
NEC will also explore support for digital securities alongside crypto assets, helping institutions develop new digital asset services.
That’s what onchain core banking is for.
📖 Read more: dfns.co/article/partne…
⌨️ Start building: app.dfns.io/get-started
NEC has spent more than 125 years building it with financial institutions.
Today, @NEC_jp and @dfnsHQ are partnering to explore digital asset infrastructure for Japanese financial institutions 🤝
The two companies have signed a ding to explore how DFNS can fit into NEC’s systems, including its planned custody offering.
The goal is to help Japanese institutions adopt digital assets without compromising their security, compliance, data residency or existing systems.
The partnership will focus on three areas:
1️⃣ Integrating DFNS wallets, key management, policies and transaction management into NEC’s digital asset infrastructure.
2️⃣ Adapting the platform to Japanese regulations and each institution’s security and operating requirements.
3️⃣ Supporting different deployment models, from SaaS to hybrid and fully on-premises, including air-gapped environments.
Institutions will be able to keep their keys, data and operations in Japan, in line with their requirements.
NEC will also explore support for digital securities alongside crypto assets, helping institutions develop new digital asset services.
That’s what onchain core banking is for.
📖 Read more: dfns.co/article/partne…
⌨️ Start building: app.dfns.io/get-started
1
2
4
202
What a beautiful day in Lisbon!
Blessed to work with partners like @Nasdaq.
Thank you @HaglindMagnus @adenatfriedman and team 🤝
Blessed to work with partners like @Nasdaq.
Thank you @HaglindMagnus @adenatfriedman and team 🤝
0
0
2
47
Welcome aboard, sir!
📈 He scaled a core banking platform to nearly 1M transactions a day.
Now he's building @dfnsHQ in Singapore.
Please welcome Zakarya Ali, our first software engineer in APAC.
Zakarya has spent more than ten years building systems where money, data and reliability meet:
→ At he ran core banking engineering. He lifted payment success and designed a double-entry ledger all under PCI and SOC 2
→ At Thales in Singapore, he led data and machine learning work for SMRT, Singapore's transit operator, and later led AI implementation for 15 engineers across Paris and Singapore
→ In six months there, he architected five production AI products, including a stablecoin payment agent, and an agentic system that cut its own running cost by 41%
→ He began at Accenture in Paris, building for clients such as Orange, La Banque Postale and EDF
Few engineers have worked on both sides of what DFNS does: the core banking systems that keep an institution's books, and the modern platforms that move value at scale. Zakarya has.
In Singapore, he'll work across our core platform and alongside clients in the region, helping institutions design and deploy on DFNS, and helping us build our engineering team in Asia.
Welcome, Zak. 🤝
📖 Read the announcement: dfns.co/article/welcom…
📮 See our open positions: jobs.deel.com/dfns
1
0
1
95
Christopher Grilhault des Fontaines retweeted
📼 Watch a transfer to an OFAC-sanctioned address get stopped.
Our CPO Chris Sutton demos the Elliptic KYT integration on @dfnsHQ, in under two minutes.
In the demo, Chris:
1️⃣ Confirms the @elliptic account on the new Integrations Hub card
2️⃣ Creates a wallet usage policy with on prescreening with Elliptic
3️⃣ Sets a risk score threshold of five, and the action to block
4️⃣ Attempts a small ETH transfer to an actual OFAC-sanctioned address, signs with a passkey
🚫 Blocked by policy. The transfer details show Elliptic's pre-screening (risk score: 10, threshold: 5)
That's the whole point of screening in the execution path rather than after settlement.
Compliance that runs after the fact is a report. Compliance that runs before the signature is a control.
And it's your risk model doing the deciding. Set the score threshold your compliance team tuned in Elliptic, name the rule categories that trigger regardless of score, choose block or request approval.
The risk appetite you defined becomes enforcement behind the API. With the Elliptic assessment attached to every decision and persisted per transfer, the proof is the record whenever the examiner asks.
That's what onchain core banking is for.
🔘 Watch the demo: youtu.be/OSCyvoa0J-8
📖 Read the announcement: dfns.co/article/ellipt…
⌨️ Start building today: app.dfns.io/get-started
Our CPO Chris Sutton demos the Elliptic KYT integration on @dfnsHQ, in under two minutes.
In the demo, Chris:
1️⃣ Confirms the @elliptic account on the new Integrations Hub card
2️⃣ Creates a wallet usage policy with on prescreening with Elliptic
3️⃣ Sets a risk score threshold of five, and the action to block
4️⃣ Attempts a small ETH transfer to an actual OFAC-sanctioned address, signs with a passkey
🚫 Blocked by policy. The transfer details show Elliptic's pre-screening (risk score: 10, threshold: 5)
That's the whole point of screening in the execution path rather than after settlement.
Compliance that runs after the fact is a report. Compliance that runs before the signature is a control.
And it's your risk model doing the deciding. Set the score threshold your compliance team tuned in Elliptic, name the rule categories that trigger regardless of score, choose block or request approval.
The risk appetite you defined becomes enforcement behind the API. With the Elliptic assessment attached to every decision and persisted per transfer, the proof is the record whenever the examiner asks.
That's what onchain core banking is for.
🔘 Watch the demo: youtu.be/OSCyvoa0J-8
📖 Read the announcement: dfns.co/article/ellipt…
⌨️ Start building today: app.dfns.io/get-started
1
1
3
264
Done ✅
Creating a million wallets used to mean a million key ceremonies.
Now it takes one API call ✔️
Today, we’re launching 𝗕𝘂𝗹𝗸 𝗔𝗱𝗱𝗿𝗲𝘀𝘀𝗲𝘀 on @dfnsHQ.
Create up to one million custodial wallets in a single request, each with its own address, name, and tags.
Creating PC takes time. That’s a problem when a payments platform needs deposit addresses for millions of users before launch.
Bulk Addresses solves this by running one MPC key ceremony per batch, then deriving each wallet address locally. No additional ceremony per wallet. No onchain activity until funds arrive.
Here’s how it works:
1️⃣ Create a batch with one API call.
2️⃣ Track progress as wallets are created in the background.
3️⃣ Retrieve the results in pages of 2,000 wallets.
Every wallet is ready to receive funds and use standard DFNS features, including webhooks and policies.
For payments, trading, and banking applications in need to swiftly create millions of deposit addresses, this makes a big difference.
Available at launch for Ethereum and its L2s, Tron, and Solana 🟢
⚠️ Bulk Addresses only works for org-controlled (“custodial”) wallets. Each batch shares a master key that cannot be exported, and the design prevents one wallet’s key from being used to reconstruct another’s.
📖 Read more: dfns.co/article/bulk-a…
⌨️ Start building: app.dfns.io/get-started
0
0
0
36
Christopher Grilhault des Fontaines retweeted
Creating a million wallets used to mean a million key ceremonies.
Now it takes one API call ✔️
Today, we’re launching 𝗕𝘂𝗹𝗸 𝗔𝗱𝗱𝗿𝗲𝘀𝘀𝗲𝘀 on @dfnsHQ.
Create up to one million custodial wallets in a single request, each with its own address, name, and tags.
Creating PC takes time. That’s a problem when a payments platform needs deposit addresses for millions of users before launch.
Bulk Addresses solves this by running one MPC key ceremony per batch, then deriving each wallet address locally. No additional ceremony per wallet. No onchain activity until funds arrive.
Here’s how it works:
1️⃣ Create a batch with one API call.
2️⃣ Track progress as wallets are created in the background.
3️⃣ Retrieve the results in pages of 2,000 wallets.
Every wallet is ready to receive funds and use standard DFNS features, including webhooks and policies.
For payments, trading, and banking applications in need to swiftly create millions of deposit addresses, this makes a big difference.
Available at launch for Ethereum and its L2s, Tron, and Solana 🟢
⚠️ Bulk Addresses only works for org-controlled (“custodial”) wallets. Each batch shares a master key that cannot be exported, and the design prevents one wallet’s key from being used to reconstruct another’s.
📖 Read more: dfns.co/article/bulk-a…
⌨️ Start building: app.dfns.io/get-started
Now it takes one API call ✔️
Today, we’re launching 𝗕𝘂𝗹𝗸 𝗔𝗱𝗱𝗿𝗲𝘀𝘀𝗲𝘀 on @dfnsHQ.
Create up to one million custodial wallets in a single request, each with its own address, name, and tags.
Creating PC takes time. That’s a problem when a payments platform needs deposit addresses for millions of users before launch.
Bulk Addresses solves this by running one MPC key ceremony per batch, then deriving each wallet address locally. No additional ceremony per wallet. No onchain activity until funds arrive.
Here’s how it works:
1️⃣ Create a batch with one API call.
2️⃣ Track progress as wallets are created in the background.
3️⃣ Retrieve the results in pages of 2,000 wallets.
Every wallet is ready to receive funds and use standard DFNS features, including webhooks and policies.
For payments, trading, and banking applications in need to swiftly create millions of deposit addresses, this makes a big difference.
Available at launch for Ethereum and its L2s, Tron, and Solana 🟢
⚠️ Bulk Addresses only works for org-controlled (“custodial”) wallets. Each batch shares a master key that cannot be exported, and the design prevents one wallet’s key from being used to reconstruct another’s.
📖 Read more: dfns.co/article/bulk-a…
⌨️ Start building: app.dfns.io/get-started
2
2
6
332
Done ✅
The largest onchain order book now has its EVM.
@dfnsHQ now supports HyperEVM, the blockchain developed by @HyperliquidX ✔️
HyperEVM brings smart contracts to the Hyperliquid ecosystem. It shares the same validators, blocks and tokens as HyperCore, the order book behind one of decentralized perpetuals.
The key difference? Smart contracts can access live order book data, including prices and positions, and send instructions back without relying on an external oracle. This opens the door to lending markets that use live positions for liquidations and products that settle directly against order book data.
More than 170 projects were building on HyperEVM by early 2026, and Tether’s USDT0 has brought significant stablecoin liquidity to the network.
With DFNS, institutions can:
1️⃣ Create wallets and hold or transfer HYPE, USDT0 and other ERC-20 tokens.
2️⃣ Interact with HyperEVM smart contracts using their ABI.
3️⃣ Apply approval workflows, transaction limits and screening before signing.
4️⃣ Sign securely and track transactions through webhooks and a single audit trail.
Trading firms can manage assets across HyperCore and HyperEVM. Treasury teams can put stablecoins to work with clear controls. Platforms can create and manage wallets for their users at scale.
HyperEVM joins more than 100 networks supported by DFNS 🟢
📖 Read more: dfns.co/article/hypere…
⌨️ Start building: app.dfns.io/get-started
0
0
0
32
Done ✅
Your own blockchain no longer needs a software release. Just an API call.
Today, we’re launching 𝗕𝗿𝗶𝗻𝗴 𝗬𝗼𝘂𝗿 𝗢𝘄𝗻 𝗖𝗵𝗮𝗶𝗻 on @dfnsHQ.
Banks and platforms increasingly run their own chains. Private Besu networks, consortium blockchains and appchains built on trum Orbit, @avax, @0xPolygon, or @CantonNetwork.
Until now, adding one meant a code change and a new release, even for on-premises deployments.
Not anymore ✔️
Register a network through the API, and it works like any of the 100+ networks supported by DFNS.
1️⃣ Create wallets with the same key across compatible chains.
2️⃣ Sign, broadcast and track transactions.
3️⃣ Index blocks, decode contract events and monitor addresses.
4️⃣ Add networks on-prem without waiting for a DFNS release.
Each network is isolated by organization. RPC credentials stay encrypted, and retired network names cannot be reused.
We started with EVM chains. More chain types to follow.
Your chain. Your configuration. Ready to use 🟢
📖 Read more: dfns.co/article/bring-…
⌨️ Start building: app.dfns.io/get-started
0
0
0
39
Christopher Grilhault des Fontaines retweeted
You give your Base address, but they send funds on Ethereum...
Same address, but wrong chain. Now what?
Our CPO Chris Sutton demos a simple fix:
➡️ Re-index a transaction by hash and network, directly from DFNS.
EVM addresses are the same across chains. Funds can arrive on a cking for that wallet.
In the demo, Chris:
1️⃣ Sends 20 USDC to his Base address on Ethereum Sepolia.
2️⃣ Creates an Ethereum Sepolia wallet for the same key, but sees a zero balance.
3️⃣ Re-indexes the transaction from organization settings.
✅ Returns to the wallet. The 20 USDC are there.
No support ticket. No waiting. Just re-index the transaction and move on.
A small feature that makes a real difference when you run digital asset operations at scale.
That's what onchain core banking is for.
🔘 Watch the demo: youtu.be/yn4KTZO3Rro
⌨️ Start building today: app.dfns.io/get-started
Same address, but wrong chain. Now what?
Our CPO Chris Sutton demos a simple fix:
➡️ Re-index a transaction by hash and network, directly from DFNS.
EVM addresses are the same across chains. Funds can arrive on a cking for that wallet.
In the demo, Chris:
1️⃣ Sends 20 USDC to his Base address on Ethereum Sepolia.
2️⃣ Creates an Ethereum Sepolia wallet for the same key, but sees a zero balance.
3️⃣ Re-indexes the transaction from organization settings.
✅ Returns to the wallet. The 20 USDC are there.
No support ticket. No waiting. Just re-index the transaction and move on.
A small feature that makes a real difference when you run digital asset operations at scale.
That's what onchain core banking is for.
🔘 Watch the demo: youtu.be/yn4KTZO3Rro
⌨️ Start building today: app.dfns.io/get-started
2
1
4
333
Christopher Grilhault des Fontaines retweeted
.@christophergdf, CEO of @dfnsHQ, says that by next year, you’ll be using tokens without even realizing it.
“It’s gonna be kinda normal to have an account, a wallet, etcetera, and it’s gonna be normal to send and receive tokens.”
“It’s gonna be kinda normal to have an account, a wallet, etcetera, and it’s gonna be normal to send and receive tokens.”
0
3
14
637
Done ✅
Until now, if we missed a deposit, you had to file a ticket and wait...
That changes today. You can now replay transactions yourself from @dfnsHQ 🟢
Replay any missed transaction by hash, from the dashboard or API. Available to every customer, at no extra cost.
Indexers can drops a block, a chain reorganizes, or a wallet is added after a deposit arrives. When that happens, you need to deal with missing balances, uncredited users, incomplete records, and more.
Now you can fix it yourself:
✔️ Paste the transaction hash in the dashboard or API.
✔️ DFNS re-fetches the transaction and updates history and balances within a minute.
✔️ Standard webhooks trigger your existing deposit and sweep workflows.
✔️ Replays are idempotent: no duplicate records or webhooks.
✔️ Access is limited to your own wallets and rate-limited.
No support ticket. No waiting. Just a complete transaction record.
It's a small feature that makes a big difference in support, speed, and reliability.
That's what onchain core banking is for.
📖 Read more: dfns.co/article/transa…
⌨️ Start building: app.dfns.io/get-started
0
0
1
48
Christopher Grilhault des Fontaines retweeted
🛡️ A certification is a promise. A surveillance audit checks you still keep it.
In August, @dfnsHQ passed its first ISO/IEC 27001, 27017 and 27018 surveillance audit, with a positive recommendation from MSECB on all three.
Over four days, auditors reviewed our security controls ises deployments, from risk management and disaster recovery to backups, incident response and employee training.
The result: all three certifications maintained. No major findings. One minor documentation issue, now corrected and closed ✅
This follows another year of work: SOC 2 renewed with no exceptions, new security controls, better monitoring and recovery, and a detailed analysis of the limits every provider faces.
Security isn’t a certificate on a wall. It’s the controls you run, test and improve every day.
The full audit report is available to clients under NDA.
⏩ Next: ISO 22301 for business continuity and SOC 1 for financial controls.
📖 Read more: dfns.co/article/iso-27…
⌨️ Start building: app.dfns.io/get-started
In August, @dfnsHQ passed its first ISO/IEC 27001, 27017 and 27018 surveillance audit, with a positive recommendation from MSECB on all three.
Over four days, auditors reviewed our security controls ises deployments, from risk management and disaster recovery to backups, incident response and employee training.
The result: all three certifications maintained. No major findings. One minor documentation issue, now corrected and closed ✅
This follows another year of work: SOC 2 renewed with no exceptions, new security controls, better monitoring and recovery, and a detailed analysis of the limits every provider faces.
Security isn’t a certificate on a wall. It’s the controls you run, test and improve every day.
The full audit report is available to clients under NDA.
⏩ Next: ISO 22301 for business continuity and SOC 1 for financial controls.
📖 Read more: dfns.co/article/iso-27…
⌨️ Start building: app.dfns.io/get-started
0
2
6
455
Christopher Grilhault des Fontaines retweeted
“A good business to be in is physical custody.”
Because every tokenized asset is backed by something real, someone has to keep it “in a secure location” and certify that it’s there.
Says, @christophergdf, CEO of @dfnsHQ on the live show.
Because every tokenized asset is backed by something real, someone has to keep it “in a secure location” and certify that it’s there.
Says, @christophergdf, CEO of @dfnsHQ on the live show.
2
1
11
679
Christopher Grilhault des Fontaines retweeted
Usually, a margin call is a lock increase.
Now it's only one atomic change, not an unlock-and-relock anymore.
Our CPO Chris Sutton demos our latest upgrade to Vaults, enabling you to modify a locked position.
In the demo, Chris:
1️⃣ Opens a vault with 5 USDC locked and scrolls hanges the locked amount from 5 to 10 USDC
3️⃣ Signs with a passkey.
✅ Locked balance: 10. Available: 10. One operation, no intermediate state
The alternative to atomic changes is to release the lock and create a new one, which opens a window where the collateral is briefly free. For a lender raising a margin requirement, or a venue extending a trading line, 𝘵𝘩𝘢𝘵 𝘸𝘪𝘯𝘥𝘰𝘸 𝘪𝘴 𝘵𝘩𝘦 𝘸𝘩𝘰𝘭𝘦 𝘳𝘪𝘴𝘬.
Modifying in place means the funds are never unlocked on the way to being locked for more. The commitment only ever moves in one step, and the counterparty's guarantee never lapses.
The same permission model holds throughout. Increases route through the committing party's approval, so a lock can't be enlarged unilaterally, and only the lock owner can reduce, release, or settle against it. Nothing locks without you, nothing unlocks without them, and now nothing changes size without both.
Off-exchange settlement, collateral, escrow. The same primitive, now adjustable without ever letting go.
That's what onchain core banking is for.
🔘 Watch the demo: youtu.be/XRrw6UV5Gtw
📖 Read it: dfns.co/article/locks
⌨️ Start building today: app.dfns.io/get-started
Now it's only one atomic change, not an unlock-and-relock anymore.
Our CPO Chris Sutton demos our latest upgrade to Vaults, enabling you to modify a locked position.
In the demo, Chris:
1️⃣ Opens a vault with 5 USDC locked and scrolls hanges the locked amount from 5 to 10 USDC
3️⃣ Signs with a passkey.
✅ Locked balance: 10. Available: 10. One operation, no intermediate state
The alternative to atomic changes is to release the lock and create a new one, which opens a window where the collateral is briefly free. For a lender raising a margin requirement, or a venue extending a trading line, 𝘵𝘩𝘢𝘵 𝘸𝘪𝘯𝘥𝘰𝘸 𝘪𝘴 𝘵𝘩𝘦 𝘸𝘩𝘰𝘭𝘦 𝘳𝘪𝘴𝘬.
Modifying in place means the funds are never unlocked on the way to being locked for more. The commitment only ever moves in one step, and the counterparty's guarantee never lapses.
The same permission model holds throughout. Increases route through the committing party's approval, so a lock can't be enlarged unilaterally, and only the lock owner can reduce, release, or settle against it. Nothing locks without you, nothing unlocks without them, and now nothing changes size without both.
Off-exchange settlement, collateral, escrow. The same primitive, now adjustable without ever letting go.
That's what onchain core banking is for.
🔘 Watch the demo: youtu.be/XRrw6UV5Gtw
📖 Read it: dfns.co/article/locks
⌨️ Start building today: app.dfns.io/get-started
1
1
6
349
Done ✅
👥 Most access controls don't fail in a breach. They fail quietly through drift.
On most platforms, approval policies name specific people. So when someone joins or leaves a group, like the “treasury team,” an admin must manually update every policy that mentions them. Do that up with dozens of hand-maintained user lists that no one fully trusts.
Today we're introducing 𝗨𝘀𝗲𝗿 𝗚𝗿𝗼𝘂𝗽𝘀, so that never has to happen again.
A User Group is a named team the platform understands.
✔️ Assign permissions to the group and every member inherits them.
✔️ Name the group as the approver and the quorum is drawn from that team.
✔️ Someone joins? Add them to the group once.
✔️ Someone leaves? Remove them once. Every policy and permission self-updates.
You can create a group with the same name as the one in your identity provider, and from there we follow your IdP (e.g. Okta, Microsoft) at every login. A match adds the user, an absence removes them, including for admin access.
Because a group is a way to grant power, it's also governed like one:
→ Adding someone to a privileged group grants the same rights as assigning them directly
→ Deleted group names stay reserved and can’t be recreated with different rights
→ A floor guard prevents removing the last administrator, directly or through group deletion
→ Every membership change is signed, audited, and traceable
The org chart and the platform now say the same thing, and your auditors can prove it.
That's what onchain core banking is for.
📖 Read it: dfns.co/article/user-g…
⌨️ Start building today: app.dfns.io/get-started
1
0
0
62
Christopher Grilhault des Fontaines retweeted
A whole new world is being built in how banks move money.
Banks will soon settle payments with tokens while the underlying deposits stay on their books.
JPMorgan is already doing it with JPM Coin, its deposit token on Base.
“Banks and [payment companies], they will start n-first operations."
"The token will move, the underlying asset will sleep, and that is a whole new world.” — @christophergdf, CEO of @dfnsHQ
Banks will soon settle payments with tokens while the underlying deposits stay on their books.
JPMorgan is already doing it with JPM Coin, its deposit token on Base.
“Banks and [payment companies], they will start n-first operations."
"The token will move, the underlying asset will sleep, and that is a whole new world.” — @christophergdf, CEO of @dfnsHQ
1
2
14
684
🚨 A must-read.
On September 24, @bitget lost $387.5 million. But no key was stolen.
The attacker took control of the server that told the signer what to sign, and the signer obeyed.
Our CISO, Thibault de Lachèze-Murel, breaks down the Bitget hack:
𝘞𝘩𝘢𝘵 𝘩𝘢𝘱𝘱𝘦𝘯𝘦𝘥, 𝘸𝘩𝘪𝘤𝘩 𝘰 𝘤𝘩𝘢𝘯𝘨𝘦, 𝘢𝘯𝘥 𝘸𝘩𝘢𝘵 𝘳𝘪𝘴𝘬𝘴 𝘳𝘦𝘮𝘢𝘪𝘯 𝘪𝘯 𝘰𝘶𝘳 𝘰𝘸𝘯 𝘮𝘰𝘥𝘦𝘭.
Based on interim reports from @Mandiant and @SlowMist_Team, here’s what happened:
→ A flaw in a third-party security product gave the attacker access 24 days before the theft.
→ A reused employee identity opened the way to the wallet server.
→ A withdrawal tool bypassed its own risk checks.
→ Bitget’s wallets signed the transfers as legitimate withdrawals.
→ $203 million moved across five chains in nine seconds.
→ Signing continued for 2h39 after the alert. Bitget stopped customer withdrawals, not the attacker.
The problem wasn’t key storage. It was who the signer trusted.
The signer checked where the request came from, not whether it was authorized. It had no proof that a person approved the transfer or that the destination and amount matched the approved instruction.
You can reduce this risk by:
✔️ Checking authority, not just origin.
✔️ Separating policy decisions from systems an attacker could compromise.
✔️ Requiring a client-controlled credential.
✔️ Limiting the warm tier with allowlists, quorums and velocity limits.
✔️ Practising how to stop signing independently of customer withdrawals.
But no vendor can eliminate every risk. Our analysis explains what providers can prevent, what clients must control, and what risks remain.
🛡️ Questions? security@dfns.co
📖 Read the analysis: dfns.co/article/the-bi…
⌨️ Start building: app.dfns.io/get-started
0
0
0
29
Christopher Grilhault des Fontaines retweeted
On September 24, @bitget lost $387.5 million. But no key was stolen.
The attacker took control of the server that told the signer what to sign, and the signer obeyed.
Our CISO, Thibault de Lachèze-Murel, breaks down the Bitget hack:
𝘞𝘩𝘢𝘵 𝘩𝘢𝘱𝘱𝘦𝘯𝘦𝘥, 𝘸𝘩𝘪𝘤𝘩 𝘰 𝘤𝘩𝘢𝘯𝘨𝘦, 𝘢𝘯𝘥 𝘸𝘩𝘢𝘵 𝘳𝘪𝘴𝘬𝘴 𝘳𝘦𝘮𝘢𝘪𝘯 𝘪𝘯 𝘰𝘶𝘳 𝘰𝘸𝘯 𝘮𝘰𝘥𝘦𝘭.
Based on interim reports from @Mandiant and @SlowMist_Team, here’s what happened:
→ A flaw in a third-party security product gave the attacker access 24 days before the theft.
→ A reused employee identity opened the way to the wallet server.
→ A withdrawal tool bypassed its own risk checks.
→ Bitget’s wallets signed the transfers as legitimate withdrawals.
→ $203 million moved across five chains in nine seconds.
→ Signing continued for 2h39 after the alert. Bitget stopped customer withdrawals, not the attacker.
The problem wasn’t key storage. It was who the signer trusted.
The signer checked where the request came from, not whether it was authorized. It had no proof that a person approved the transfer or that the destination and amount matched the approved instruction.
You can reduce this risk by:
✔️ Checking authority, not just origin.
✔️ Separating policy decisions from systems an attacker could compromise.
✔️ Requiring a client-controlled credential.
✔️ Limiting the warm tier with allowlists, quorums and velocity limits.
✔️ Practising how to stop signing independently of customer withdrawals.
But no vendor can eliminate every risk. Our analysis explains what providers can prevent, what clients must control, and what risks remain.
🛡️ Questions? security@dfns.co
📖 Read the analysis: dfns.co/article/the-bi…
⌨️ Start building: app.dfns.io/get-started
The attacker took control of the server that told the signer what to sign, and the signer obeyed.
Our CISO, Thibault de Lachèze-Murel, breaks down the Bitget hack:
𝘞𝘩𝘢𝘵 𝘩𝘢𝘱𝘱𝘦𝘯𝘦𝘥, 𝘸𝘩𝘪𝘤𝘩 𝘰 𝘤𝘩𝘢𝘯𝘨𝘦, 𝘢𝘯𝘥 𝘸𝘩𝘢𝘵 𝘳𝘪𝘴𝘬𝘴 𝘳𝘦𝘮𝘢𝘪𝘯 𝘪𝘯 𝘰𝘶𝘳 𝘰𝘸𝘯 𝘮𝘰𝘥𝘦𝘭.
Based on interim reports from @Mandiant and @SlowMist_Team, here’s what happened:
→ A flaw in a third-party security product gave the attacker access 24 days before the theft.
→ A reused employee identity opened the way to the wallet server.
→ A withdrawal tool bypassed its own risk checks.
→ Bitget’s wallets signed the transfers as legitimate withdrawals.
→ $203 million moved across five chains in nine seconds.
→ Signing continued for 2h39 after the alert. Bitget stopped customer withdrawals, not the attacker.
The problem wasn’t key storage. It was who the signer trusted.
The signer checked where the request came from, not whether it was authorized. It had no proof that a person approved the transfer or that the destination and amount matched the approved instruction.
You can reduce this risk by:
✔️ Checking authority, not just origin.
✔️ Separating policy decisions from systems an attacker could compromise.
✔️ Requiring a client-controlled credential.
✔️ Limiting the warm tier with allowlists, quorums and velocity limits.
✔️ Practising how to stop signing independently of customer withdrawals.
But no vendor can eliminate every risk. Our analysis explains what providers can prevent, what clients must control, and what risks remain.
🛡️ Questions? security@dfns.co
📖 Read the analysis: dfns.co/article/the-bi…
⌨️ Start building: app.dfns.io/get-started
0
4
16
1,091
I'm joining @stabledash Live today at 3:00 PM ET to make the case that @dfnsHQ will always be a pure technology service provider.
x.com/i/broadcasts/1…
x.com/i/broadcasts/1…
0
0
1
41