The week in breach news Nov 19, 2025

Anthropic, a San Francisco-based artificial intelligence (AI) safety and research company, claims to have stopped what it believes is the first documented large-scale cyberattack carried out primarily by AI. According to the company, an estimated 80–90% of the work behind this attack was executed by AI.

Anthropic reported on November 13 that a Chinese state-sponsored group manipulated its AI-powered coding assistant, Claude Code, and attempted to infiltrate roughly 30 global targets. The targets included major tech firms, financial institutions, chemical manufacturers and government agencies. The company said it detected suspicious activity in mid-September, which later proved to be part of a highly sophisticated espionage campaign. To bypass safety controls, the attackers posed as a legitimate cybersecurity firm performing defensive testing and jailbroke Claude Code, pushing it beyond its guardrails.

Reports indicate that the threat actors used AI's agentic capabilities at a level not previously seen. The incident is being viewed as a turning point, one where AI is no longer limited to advising attackers but is now being used to execute key stages of a cyberattack itself.

In yet another cyberattack targeting U.S. universities, Princeton University has reported a breach involving a database containing information about alumni, donors, students and other community members.

According to the university, the breach occurred on November 10 and involved a database managed by the University Advancement office. Princeton sent notification emails on November 15 to individuals who may have been affected. The compromised database did not contain Social Security numbers, passwords or financial information such as credit card or bank account details. However, it did include personal information such as names, email addresses, phone numbers, home and business addresses and donation histories.

This marks another major cyber incident in higher education this month. Earlier in November, the University of Pennsylvania reported a breach that exposed more than 1.2 million records.

On November 10, about 200,000 New York residents received scam texts after the vendor behind the state's official text-messaging system was compromised.

Mobile Commons, a company used by local and state governments to send public service alerts, was hacked last week, and its platform was exploited to disseminate fraudulent texts. The company stated that an unauthorized third party gained access to its system on Monday through a spear-phishing attack or a similar social engineering method. During the four hours it took to shut down the activity, residents received scam messages urging them to call a toll-free number about a declined bank transaction involving a large sum of money — an attempt to trick victims into completing a real transaction.

This latest campaign highlights how quickly scam-text operations are escalating. It remains unclear how many people responded to the messages or suffered financial losses.

Miniatur Wunderland, a popular miniature museum in Hamburg, Germany, reported a data breach that exposed the credit card details of hundreds of thousands of visitors.

In an email to affected customers, the museum confirmed that unauthorized third parties had compromised its online ticketing system, giving them access to sensitive payment information. The notice explained that anyone who purchased tickets online between June 6 and October 29 may have had their credit card data exposed. The breach may have exposed the cardholder's name, card number, expiration date and CVV security code.

While more than 1.5 million people visit the museum each year, only customers who bought tickets online during the affected five-month period were impacted.

The Qilin ransomware group allegedly breached Cornerstone Staffing Solutions, a U.S.-based recruitment agency and HR technology company and claims to have stolen personal resumes belonging to more than 120,000 job seekers.

On November 13, Qilin claimed in a post on its dark web leak site that it had compromised the agency and exfiltrated 300 GB of sensitive information. The group stated that the stolen resumes were part of a larger cache of nearly 1 million files, containing nearly 24 million pieces of personal information, according to the proof samples shared on its website.

Qilin has threatened to release the data publicly if Cornerstone does not begin negotiations.