Daily Dose of Cybersecurity News - September 1, 2025

What happened

AIDEFEND, an open-source framework dedicated to AI security, has been released to provide defensive countermeasures and best practices for safeguarding AI and machine learning systems.

Who is affected

Security professionals and organizations utilizing AI and machine learning technologies.

Why it matters

As AI systems become integral to various applications, they are increasingly targeted by adversaries. AIDEFEND offers actionable strategies to protect these systems from emerging threats.

How it could have been prevented

Implementing comprehensive security frameworks like AIDEFEND can proactively address vulnerabilities in AI systems, reducing the risk of exploitation.

Recommended reading: LlamaFirewall: Open-source framework to detect and mitigate AI centric security risks

What happened

Google Cloud's Office of the CISO released a report emphasizing the need for boards of directors to take a more active role in cybersecurity oversight, focusing on ransomware, cyber-enabled fraud, and the integration of security in innovation processes.

Who is affected

Boards of directors and executive leadership teams across various industries are called upon to enhance their engagement in cybersecurity governance.

Why it matters

As cyber threats become more sophisticated, board-level involvement is crucial to ensure comprehensive risk management, protect organizational assets, and maintain customer trust.

How it could have been prevented

Boards should proactively engage in cybersecurity strategy, ensure the implementation of robust identity protection measures, and integrate security considerations into all stages of product and service development.

Recommended reading: Balancing cybersecurity with business priorities: Advice for Boards

What happened

Over 80% of large U.S. companies experienced socially engineered fraud in the past year, with nearly half incurring financial losses exceeding $500,000. These attacks often exploited gaps between finance and security teams, leveraging generative AI to execute complex schemes.

Who is affected

Large U.S. enterprises across various sectors, particularly those with misaligned finance and security operations.

Why it matters

The integration of generative AI into fraud tactics has made attacks more sophisticated and harder to detect, leading to significant financial and reputational damage. Operational silos between departments exacerbate these vulnerabilities.

How it could have been prevented

Enhancing collaboration between finance and security teams to ensure unified visibility and response to fraud incidents. Implementing AI-driven behavioral monitoring systems to detect anomalies across multiple platforms.

Recommended reading: How cybercriminals are weaponizing AI and what CISOs should do about it

What happened

Over 300,000 internet-facing Plex Media Server instances remain vulnerable to CVE-2025-34158, a critical flaw allowing remote code execution. Additionally, CVE-2025-48384, a Git vulnerability enabling arbitrary file writes and remote code execution, is being actively exploited.

Who is affected

Users and organizations running outdated versions of Plex Media Server (versions 1.41.7.x to 1.42.0.x) and Git clients prior to versions 2.43.7 through 2.50.1.

Why it matters

Exploitation of these vulnerabilities can lead to unauthorized access, data breaches, and full system compromise, posing significant risks to personal and organizational data security.

How it could have been prevented

Regularly updating software to the latest versions and applying security patches promptly can mitigate such vulnerabilities. For Git, avoiding recursive submodule clones from untrusted sources and disabling Git hooks globally can reduce risk.

Recommended reading: Plex Media Server CVE-2025-34158 Attack

Recommended reading: Git Vulnerability Exploited CVE-2025-48384