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:
- Update now! Atlassian Confluence vulnerability is being actively exploited
October 12, 2023
Microsoft Threat Intelligence has revealed that it has been tracking the active exploitation of a vulnerability in Atlassian Confluence software since September 14, 2023. At the time the attacks were first observed the vulnerability was a zero-day, meaning that no update was available, so defenders had “zero days” to patch the flaw. The vulnerability has since ...
- Akira ransomware overview
October 12, 2023
Akira is a relatively new ransomware variant with Windows and Linux versions that came out in April 2023. Like many attackers, the gang behind this variant only uses the ransomware to encrypt files after first breaking into a network and stealing data. This group also employs a double extortion tactic, demanding a ransom from victims ...
- CISA Releases Nineteen Industrial Control Systems Advisories
October 12, 2023
CISA released nineteen Industrial Control Systems (ICS) advisories on October 12, 2023. These advisories provide timely information about current security issues, vulnerabilities, and exploits surrounding ICS. ICSA-23-285-01 Siemens SIMATIC CP products ICSA-23-285-02 Siemens SCALANCE W1750D ICSA-23-285-03 Siemens SICAM A8000 Devices Read more… Source: U.S. Cybersecurity and Infrastructure Security Agency
- ToddyCat: Keep calm and check logs
October 12, 2023
ToddyCat is an advanced APT actor that Kaspersky researchers described in a previous publication last year. The group started its activities in December 2020 and has been responsible for multiple sets of attacks against high-profile entities in Europe and Asia. Kaspersky first publication was focused on their main tools, Ninja Trojan and Samurai Backdoor, and ...
- India’s Bank of Baroda expose worsens: Agents steal money from accounts
October 12, 2023
India’s Bank of Baroda made it simple and easy for its agents to steal money from the accounts of its customers. And some of them did steal 2.2 million rupees ($27,000) from 362 customers, internal audit reports and records of the bank have revealed. The audits come after an expose by The Reporters’ Collective (TRC) and ...
- Automatic disruption of human-operated attacks through containment of compromised user accounts
October 11, 2023
Based on incidents analyzed by Microsoft, it can take only a single hop from the attacker’s initial access vector to compromise domain admin-level accounts. For instance, an attacker can target an over-privileged service account configured in an outdated and vulnerable internet-facing server. Highly privileged user accounts are arguably the most important assets for attackers. Compromised domain ...