News
Autopay Says Cyberattack Was Heavily AI-Assisted, Scale of Data Exposure Unknown

Payment operator Autopay detected unauthorized access to its infrastructure and some payment data on 11 November. The company describes the attack as heavily AI-assisted, but it has not confirmed that data was copied or said how many people are affected.
Contents
Autopay, a Polish company that handles online payments, has disclosed a cyberattack that, by its own analysis, showed signs of automation heavily assisted by artificial intelligence. The scope of the incident is still under investigation, and the company has not confirmed that any data was copied or used, though it cannot rule that out.
What the company has established
Autopay detected the unauthorized access on 11 November and published a statement on Friday, 13 November. According to money.pl, citing the PAP news agency, the company blocked the access, cut off the identified channels, revoked credentials, isolated the affected servers and put additional monitoring in place.
The company is notifying individuals whose data may be involved. A description of how the attackers operated was passed to CSIRT KNF so it can be distributed across the financial sector. Autopay says it is withholding technical details for the sake of its own security and that of other market participants.
So far we have not confirmed that the data was copied or used, but we cannot rule it out entirely, so we are informing customers directly to prevent possible abuse. - Zuzanna Szybisty of Autopay's press office, on the possibility that data was copied or used
The role of AI
The company describes the attack as an entirely new type, making heavy use of AI tools that many firms are not yet prepared for. According to the analysis so far, the attackers used neither malware nor phishing.
As Spider's Web reports, citing information given to Sekurak, the attack was highly automated, and the company believes the attackers most likely used language models. After an initial phase, they reportedly accelerated sharply and spread across many resources, then changed tactics quickly when access routes were blocked.
The same account says the attackers used tools built ad hoc for the specific environment. They did not target workstations but moved through trusted systems: monitoring, CI/CD, deployment automation and password management. Much of the activity relied on legitimate credentials and administrative tools, so individual actions did not trigger typical alerts. The use of AI is so far the company's own finding, not a conclusion independently confirmed by law enforcement or regulators.
Card data: conflicting reports
The scope of the data remains unclear. CyberDefence24, citing Sekurak and the company's messages to customers, lists an email address, a phone number (only for phone top-ups), an IP address and the details of the card used for payment. The outlet stresses that a leak has not been confirmed and that it is describing a risk.
Spider's Web points out that Autopay has not said which categories of data were involved, and that there is no basis for claiming that card numbers, security codes or online banking passwords were taken. The author also warns against two extremes: downplaying the incident just because copying has not been confirmed, and announcing a leak affecting millions of Poles without evidence.
What customers should do
The company recommends regularly checking transaction history and promptly reporting any operations you did not make to the bank that issued your card. It also warns about messages and phone calls that invoke the incident. Spider's Web notes that these may be fraud attempts, for example a request for urgent account verification.
For payment and e-commerce companies, one point the company raised matters: traffic through trusted systems and legitimate credentials makes individual actions harder to detect. Autopay invites firms and institutions that want to exchange experiences to write to cybersecurity@autopay.pl. Key rotation with partners and a post-breach analysis with external support are under way, and are meant to establish what the attackers had access to and whether any data left the infrastructure.

