ISRAELI FIRM’S REPORT
Taiwan’s assertion got here a day after the Israeli cybersecurity firm Dream mentioned in a weblog publish that it had uncovered an AI-driven hacking marketing campaign that stole credentials and different information from an unnamed authorities in Asia.
Dream mentioned a group of AI brokers labored collectively to extract scores of passwords from unidentified officers, steal personnel information from Taiwan’s justice ministry, and scan its nuclear security company for vulnerabilities over the course of 4 days.
Dream, which mentioned it was capable of reconstruct the hacking marketing campaign after recovering the brokers’ “full operational workspace”, declined to share the information or title the federal government when approached by Reuters, however the Monetary Occasions, which was the primary media outlet briefed on Dream’s findings, recognized the goal businesses as Taiwanese.
GROWING THREAT
The specter of AI-assisted hacking campaigns has been rising for years however ramped up dramatically over the previous few months after high AI labs launched fashions, akin to Anthropic’s Mythos, which may quickly conduct reconnaissance, establish vulnerabilities and make the most of them.
Cris Thomas, a researcher who works as a safety advocate on the code safety firm Semgrep, mentioned the spy marketing campaign was an illustration of simply how shortly AI-assisted hackers may hit their mark.
Thomas cautioned towards exaggerating the facility of AI brokers, nevertheless.
“There’s nonetheless a human in there someplace. Any individual had to decide on who to assault, needed to set up an goal and provides it a directive. It is not completely 100 per cent autonomous. There was a succesful operator in cost that did that,” he added.
