🌍The Next Web (TNW)•Freshcollected in 3h
Fraudsters Used Stolen Funds to Buy Claude Credits
💡A real fraud case shows how chatbot credits can turn into a new payment-abuse vector.
⚡ 30-Second TL;DR
What Changed
A Sussex businessman lost £14,244 from his Metro Bank account.
Why It Matters
The case shows that AI services with prepaid credits can create new abuse and payment-fraud surfaces. AI providers and banks may need stronger transaction monitoring, credit-purchase limits, and rapid blocking workflows.
What To Do Next
Audit your AI product’s prepaid-credit flow by adding velocity limits, device and payment-risk scoring, and automatic suspension after the first confirmed fraud signal.
Who should care:Developers & AI Engineers
Key Points
- •A Sussex businessman lost £14,244 from his Metro Bank account.
- •The stolen funds were used to purchase Claude chatbot credits.
- •The victim already had a paid Claude subscription.
- •Metro Bank reportedly detected the first fraudulent charge but allowed further charges.
🧠 Deep Insight
AI-generated analysis for this event.
🔑 Enhanced Key Takeaways
- •The fraud involved the exploitation of Anthropic's API or subscription billing systems, which are increasingly targeted by 'carding' operations due to the high resale value of AI compute credits on the dark web.
- •Metro Bank faced significant public criticism for its fraud detection protocols, specifically the failure to trigger a 'stop' on the account after the initial anomalous transaction was flagged.
- •Security researchers note that AI service providers are struggling to implement robust velocity checks for credit purchases, making them attractive targets for money laundering via stolen credit cards.
- •The victim's existing subscription likely provided the fraudsters with a 'trusted' baseline, allowing them to bypass some automated fraud filters that might otherwise block sudden, high-value purchases.
- •This incident has prompted discussions among UK financial regulators regarding the liability of banks when they identify suspicious activity but fail to prevent subsequent unauthorized transactions.
📊 Competitor Analysis▸ Show
| Feature | Claude (Anthropic) | ChatGPT (OpenAI) | Gemini (Google) |
|---|---|---|---|
| Primary Model | Claude 3.5 Sonnet/Opus | GPT-4o | Gemini 1.5 Pro |
| Billing Security | Standard Stripe/API integration | Advanced fraud detection/Velocity limits | Integrated Google Pay/Cloud security |
| Enterprise Focus | High (Constitutional AI) | High (Custom GPTs) | High (Workspace Integration) |
| Credit Resale Risk | High (High demand for API) | Very High (Massive ecosystem) | Moderate (Tied to Google accounts) |
🛠️ Technical Deep Dive
- The attack likely utilized automated scripts to interact with the Anthropic billing API, bypassing standard web-based UI fraud protections.
- Fraudsters often use 'carding' bots that test stolen credentials against various services to verify validity before attempting larger transactions.
- AI credit systems often lack multi-factor authentication (MFA) for individual credit top-ups once a payment method is saved to the account.
- The vulnerability stems from the decoupling of the user's account identity from the payment authorization flow, allowing attackers to leverage existing payment tokens.
🔮 Future ImplicationsAI analysis grounded in cited sources
AI providers will mandate MFA for all credit purchases exceeding a specific threshold.
To mitigate carding risks, platforms must move beyond simple saved payment methods to verify the identity of the purchaser for high-value transactions.
Banks will implement mandatory 'transaction pauses' for all digital service subscriptions after a fraud alert.
Regulatory pressure following high-profile failures like this will force financial institutions to prioritize account freezing over transaction processing during active fraud investigations.
⏳ Timeline
2023-03
Anthropic launches Claude, introducing the subscription and API credit model.
2024-06
Anthropic releases Claude 3.5 Sonnet, significantly increasing demand for API credits.
2026-07
The reported fraudulent transactions occur on the victim's Metro Bank account.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: The Next Web (TNW) ↗