Cisco Talos

Cisco Talos researchers have identified 84 malware samples designed to manipulate large language model-based analysis systems. The samples use techniques ranging from deceptive embedded comments to exploitation of AI guardrails, aiming to fool automated security tools that rely on LLMs to inspect suspicious code. It is an early sign that attackers are adapting their methods specifically for AI-assisted cyber defence.

Cisco Talos has published its latest Threat Source newsletter, outlining emerging techniques that malware uses to evade AI-based analysis tools, alongside coverage of newly disclosed critical software vulnerabilities. The report argues attackers are adapting their methods specifically to defeat automated security research, prompting discussion among cybersecurity professionals about how defenses must evolve.

Cybersecurity firm Cisco Talos has published research on how malware authors are using AI-analysis evasion techniques to slip past automated defences. In a nod to the trend, the blog is titled with a prompt-injection style phrase, 'Ignore all instructions', the same kind of command attackers embed in malicious code to confuse AI-based scanning tools. The report examines the current state of these evasion methods and what defenders can do about them.

Cisco has released a batch of security advisories that include nine vulnerabilities rated critical. The advisories were published through Cisco's security portal and highlighted by Talos, the company's threat intelligence team. Security professionals are urging administrators to review the disclosures and apply patches promptly, as critical-rated flaws can often allow remote code execution or full system compromise if left unaddressed.