Threat Brief — 2026-08-17 — Crypto Wallet Breach, AI Outage
Executive summary: SafePal, a cryptocurrency hardware wallet provider, disclosed a data breach affecting nearly 40,000 customers after an exploitation of a flaw exposed order information now offered for sale by a threat actor. Separately, Anthropic confirmed a major outage taking Claude offline across multiple services — a reminder that AI-platform availability is now operationally critical for many engineering teams.
Top items
- SafePal data breach — 39,798 customer records for sale. A flaw was exploited to extract customer order information from the cryptocurrency hardware wallet provider; a threat actor is actively offering the stolen dataset for purchase. Customers who purchased SafePal devices should assume personal and order details are exposed and watch for targeted phishing or social-engineering attempts. (src: BleepingComputer)
- Anthropic Claude major outage — multiple services degraded. Anthropic has confirmed Claude is down in a widespread outage, with users reporting login failures and degraded performance across several services. Teams embedding Claude in production workflows should check status pages and have fallback plans. (src: BleepingComputer)
Themes
Crypto-adjacent supply chain targeting continues. The SafePal breach follows a pattern seen earlier this fortnight (Trezor customer-data breach via ShipMonk, Coinsbuy $8M heist) where attackers target the customer-data perimeter around crypto companies rather than the wallets themselves. The goal is likely credential phishing and social engineering rather than direct key theft.
AI-platform reliability is now a security concern. The Claude outage arrives alongside last week's story of Claude making an autonomous HR firing decision — reinforcing that AI services are deeply embedded in operational workflows, making their availability and failure modes a first-order risk for engineering organisations.