Microsoft has discovered a new method for jailbreaking Large Language Model (LLM) Artificial Intelligence (AI) tools and has revealed its continued efforts to improve LLM safety and security. Microsoft described the "Crescendo" LLM jailbreak method in a recent paper, delving into how an attacker can send a series of seemingly benign prompts to gradually lead a chatbot, such as OpenAI's ChatGPT, Google's Gemini, Meta's LlaMA, or Anthropic's Claude, to deliver output that the LLM model would normally filter and refuse.

A team of researchers at Pohang University of Science and Technology (POSTECH) significantly advanced online security by integrating an Artificial Intelligence (AI)-based metasurface with oblique helicoidal cholesteric liquid crystals. The continued development of digital technology calls for greater security for personal data online. Encryption systems are critical to the stability of Internet communication and serve as the foundation for online privacy.

A social engineering attack has compromised a third-party provider responsible for handling telephony for Cisco's Duo Multi-Factor Authentication (MFA) service. Cisco Duo customers have been warned to be on the lookout for follow-on phishing attacks. Customers were notified that the company handling SMS and VOIP MFA messaging traffic for Cisco Duo had been breached on April 1. The threat actor allegedly used compromised employee credentials, and once inside the service provider's systems, they downloaded SMS logs for specific users.

The OpenJS Foundation has thwarted a "credible takeover attempt" similar to the one that resulted in a backdoor being put in the open source XZ Utils package by someone called "Jia Tan." The malicious maintainer achieved that position through a successful long-term social engineering campaign. Lasse Collin, the project's author and primary maintainer, was convinced to share the duty of keeping the project running smoothly.

The threat actor known as "TA558" has been using steganography as an obfuscation method in the delivery of a variety of malware, including Agent Tesla, FormBook, Remcos RAT, LokiBot, GuLoader, Snake Keylogger, XWorm, and more. Positive Technologies reported that the group used steganography extensively, sending VBSs, PowerShell code, and RTF documents containing an embedded exploit, inside images and text files. The campaign has been dubbed "SteganoAmor" due to its use of steganography and choice of file names.

Fabian Baumer and Marcus Brinkmann from Ruhr University Bochum discovered a vulnerability in PuTTY 0.68 through 0.80 that enables attackers with access to 60 cryptographic signatures to recover the private key used to generate them. PuTTY is a popular open source terminal emulator, serial console, and network file transfer tool that supports SSH, Telnet, SCP, and SFTP. According to the researchers, the vulnerability stems from how PuTTY generates ECDSA nonces, which are temporary unique cryptographic numbers, for the NIST P-521 curve used for SSH authentication.