A single compromised web-testing tool may have exposed visitors to 178 major websites to crypto theft and payment-card skimming.
Hackers injected malicious code into Optimizely’s web testing software on Monday, adding cryptocurrency wallet addresses and hijacking payment forms across 178 organizations, including HSBC, Salesforce, the BBC, and hardware wallet maker Ledger.
Websites embed snippets of Optimizely's code to test different page layouts and copy, causing visitors' browsers to run the vendor's software automatically on each visit.
An unauthorized party exploited a flaw in Optimizely's account signup flow on Nov. 16 to insert rogue code into customer projects, the company said, adding that it removed the payload and fixed the flaw.
The attacker didn’t need to breach any of the affected sites directly. Optimizely’s software lets companies test different versions of their web pages, and each site that uses it pulls the company’s code on every visit.
The company has since removed the code and fixed the flaw.
The attack focused on sites that accept crypto payments. The code scanned pages for wallet addresses on Bitcoin, Ethereum, Solana, Tron and five other networks and replaced them with the attacker’s own, both in form fields and in text visitors copied to their clipboards. A user who pasted a substituted address into a wallet and approved the transfer would have sent funds to the attacker.
The script also skimmed card numbers entered into sites’ payment forms, copied cookies and screenshots, and served exploit kits targeting older iPhones.
Six of the organizations identified by Helme are crypto or trading businesses, including Ledger, IG Group Holdings Plc, OANDA’s Coinpass, Interactive Investor, TMS Brokers and Betashares. Another 32 are banks and financial companies.
Wallets linked to the attack had received about $7,040, mostly in Bitcoin, by Monday evening, according to Report URI. Bloomberg couldn’t independently verify the figure.
Representatives for Optimizely, Ledger and HSBC didn’t immediately respond to CoinDesk’s requests for comment.
The attacker never had to breach Ledger’s or HSBC’s own servers. Websites run whatever code their vendors send them, so a single break-in at a vendor can push malicious code to hundreds of customer sites at once. Optimizely wasn’t the only route either. Hudson Rock, a firm that tracks infostealer malware, found similar code moving through three other web services, starting with an earlier wave on Nov. 13.
Swapping addresses is an old trick in crypto theft, and attackers keep using it because they never have to break into a wallet. They only have to change the address a user copies before hitting send.
None of the firms tracing the attack have linked it to AI tools so far, and Optimizely blamed a flaw in its signup process.
The attack still lands months after AI research firm Anthropic released Claude Mythos Preview, a model it limited to a small group of trusted organizations because it finds software flaws so well. That release set off months of warnings about AI-powered hacking.
Monthly vulnerability disclosures roughly doubled between January and July, according to a Monday report by Google’s threat intelligence team.
Read More: How Anthropic’s Mythos model is forcing the crypto industry to rethink everything about security



