From August–September 2026, Chinese-speaking attackers used an open-source AI penetration testing framework called Cairn to autonomously conduct 25 operations against at least 19 organizations. The attackers weaponized Cairn by pairing it with Chinese AI models Zhipu GLM and DeepSeek.
The attackers supplied a target and a goal, then Cairn planned, executed and recorded results on its own. These attacks achieved unauthorized access or data retrieval at 10 of the targeted organizations, including:
- Accessing dozens of valid credentials
- Mass exposing personal data (exceeding one million records at one organization)
- Retrieving military and government identities
- Creating persistent administrator accounts in a production system
- Modifying a live industry standards-ballot document
Read more…
Source: Palo Alto Unit 42
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- BlueLeaks: Data from 200 US police departments & fusion centers published online
June 22, 2020
The files, dubbed BlueLeaks, have been published by Distributed Denial of Secrets (DDoSecrets), a group that describes itself as a “transparency collective.” The data has been made available online on a searchable portal. According to the BlueLeaks portal, the leaked data contains more than one million files, such as scanned documents, videos, emails, audio files, and more. DDoSecrets ...
- Microcin is here
June 19, 2020
In February 2020, we observed a Trojan injected into the system process memory on a particular host. The target turned out to be a diplomatic entity. What initially attracted our attention was the enterprise-grade API-like (application programming interface) programming style. Such an approach is not that common in the malware world and is mostly used ...
- Hackers use fake Windows error logs to hide malicious payload
June 19, 2020
Hackers have been using fake error logs to store ASCII characters disguised as hexadecimal values that decode to a malicious payload designed to prepare the ground for script-based attacks. The trick is part of a longer chain with intermediary PowerShell commands that ultimately delivers a script for reconnaissance purposes. MSP threat detection provider Huntress Labs discovered an attack scenario ...
- Australian PM says nation under serious state-run ‘cyber attack’ – Microsoft, Citrix, Telerik UI bugs ‘exploited’
June 19, 2020
Australian Prime Minister Scott Morrison has called a snap press conference to reveal that the nation is under cyber-attack by a state-based actor, but the nation’s infosec advice agency says that while the attacker has gained access to some systems it has not conducted “any disruptive or destructive activities within victim environments.” Morrison said the attack ...
- Forward-looking security analysis of smart factories [Part 4] – MES database compromises
June 18, 2020
If I had to describe the role of the MES (Manufacturing Execution System) in a single phrase, it would be “manufacturing playmaker.” As I mentioned in the previous column, the manufacturing process is divided into multiple layers, and the system performs a different role in each layer. The MES is at the top layer of the ...
- Why did this Bank of America phishing email bypass spam filters?
June 18, 2020
Threat actors trying to steal your credentials through phishing attackers is nothing new, and the number of campaigns has only been rising in recent times. Government estimates indicate that phishing is a multi-billion dollar industry, which is why cybersecurity companies exist that focus entirely on securing client’s inboxes from malicious email. Due to this, attackers continuously come up ...