Plainly speaking, Artificial intelligence is no longer a fringe technology. It has become a core component of modern business, from customer service chatbots to complex data analysis.
We often treat the Large Language Models (LLMs) that are at the core of this technology as trusted black boxes. But like any software, they can be tampered with, manipulated, and turned against their creators. Understanding the ways an AI model can be compromised is the first step toward building a secure and resilient AI infrastructure. There are three primary ways that the authors want to cover, in which they will explain how an AI model can be compromised and made to act in unpredictable (or malicious!) ways:
Read more…
Source: Trend Micro
Sign up for the Cyber Security Review Newsletter
The latest cyber security news and insights delivered right to your inbox
Related:
- The impact of legacy vulnerabilities in today’s cybersecurity landscape
June 4, 2024
Of the top five most widely used network attacks against SMBs, the ‘newest’ vulnerability represented were nearly three years old, while the oldest were over a decade old – which is primitive when considering the modern threat environment. The results are a clear reminder for CISOs and cybersecurity leaders that they must assess organizational threats based ...
- Navigating Cyber Risk Through Collaboration
June 3, 2024
The constantly evolving cyber risk landscape presents a formidable challenge to organizations, as businesses transform, scopes and boundaries shift, and bad actors develop new tactics and techniques to exploit vulnerabilities and compromise systems. With the rate of risk velocity increasing faster than the resources of most risk-focused teams, collaboration and working together smarter is one of ...
- Excel File Deploys Cobalt Strike at Ukraine
June 3, 2024
FortiGuard Labs has recently identified a sophisticated cyberattack involving an Excel file embedded with a VBA macro designed to deploy a DLL file. The attacker uses a multi-stage malware strategy to deliver the notorious “Cobalt Strike” payload and establish communication with a command and control (C2) server. This attack employs various evasion techniques to ensure successful ...
- UK: The Princess of Wales’s hospital data breach not referred to police due to suspected ‘decoy’ plan
June 3, 2024
The Princess of Wales’s hospital data breach has not been referred to police as an expert explains that a “decoy” plan could have been in use – meaning her actual medical files were not accessed by the perpetrators. Despite Health Minister Maria Caulfield revealing back in March that the police had been asked to look into ...
- Inside The Box: Malware’s New Playground
June 3, 2024
Over the past few months, we have been monitoring the increasing abuse of BoxedApp products in the wild. BoxedApp products are commercial packers that provide advanced features such as Virtual Storage (Virtual File System, Virtual Registry), Virtual Processes, and a universal instrumentation system (WIN/NT API hooking). Even though BoxedApp has been commercially available for a while, ...
- Ticketmaster confirms customer data breach
June 1, 2024
Live Nation Entertainment has confirmed what everyone has been speculating on for the last week: Ticketmaster has suffered a data breach. In a filing with the SEC, Live Nation said on May 20th it identified “unauthorized activity within a third-party cloud database environment containing Company data (primarily from its Ticketmaster L.L.C. subsidiary)” and launched an investigation. ...