Critical

Attacker Hijacks AI Coding Assistant Session, Spreads Shai-Hulud Across About 100 Repositories

The Hacker News
Actively Exploited

Overview

Mandiant has reported a significant security incident involving the hijacking of an AI coding assistant session at a software-as-a-service provider. An attacker exploited this session to introduce a malicious worm known as Shai-Hulud into approximately 100 internal code repositories. The attack unfolded when the compromised assistant recommended software that had been tampered with, and this recommendation was unwittingly accepted by users. As a result, the worm was able to steal sensitive data, including repository secrets and source code. This incident raises serious concerns about the security of AI tools in software development and the potential for similar attacks to compromise sensitive information across multiple organizations.

Key Takeaways

  • Active Exploitation: This vulnerability is being actively exploited by attackers. Immediate action is recommended.
  • Affected Systems: AI coding assistant, Shai-Hulud worm, internal code repositories
  • Action Required: Users should review and secure their AI coding assistant configurations, audit repository access controls, and monitor for any unauthorized software recommendations or activities.
  • Timeline: Newly disclosed

Original Article Summary

Mandiant says an attacker hijacked an active AI coding-assistant session at an unnamed software-as-a-service provider and later spread Shai-Hulud across about 100 internal code repositories. Before the repository spread, the assistant recommended software that the attacker had poisoned, and the recommendation was accepted. The worm stole repository secrets and source code for the

Impact

AI coding assistant, Shai-Hulud worm, internal code repositories

Exploitation Status

This vulnerability is confirmed to be actively exploited by attackers in real-world attacks. Organizations should prioritize patching or implementing workarounds immediately.

Timeline

Newly disclosed

Remediation

Users should review and secure their AI coding assistant configurations, audit repository access controls, and monitor for any unauthorized software recommendations or activities.

Additional Information

This threat intelligence is aggregated from trusted cybersecurity sources. For the most up-to-date information, technical details, and official vendor guidance, please refer to the original article linked below.

Related Topics: This incident relates to Data Breach, Malware.

Related Coverage

CISA promotes a fresh way to deter cyberattackers: Lie to them

CyberScoop

The Cybersecurity and Infrastructure Security Agency (CISA) has released its first guidance on using decoys, such as honeypots, to combat cyberattacks. This approach aims to mislead attackers by providing false information and diverting them from valuable targets. By deploying these deceptive tactics, organizations can not only detect intrusions more effectively but also buy time to respond to threats. This guidance represents a shift in how organizations can defend against cyber threats, emphasizing the need for creativity in cybersecurity strategies. As cyberattacks become more sophisticated, CISA’s recommendations may help organizations better protect their assets and data.

Sep 16, 2026

Data Broker Radaris Loses Domains in Privacy Fight

Krebs on Security

Radaris.com, a consumer data broker known for its people-search services, has faced legal repercussions for allegedly violating New Jersey privacy laws. The lawsuit centered on Radaris's failure to comply with requests to remove personal information, particularly concerning state law enforcement officials. Following the company's repeated resistance in court, a judge ruled that Radaris must transfer its domain and over a dozen related sites to the plaintiffs. This case highlights ongoing concerns about data brokers' practices and the challenges individuals face in protecting their personal information online. The outcome could set a precedent for how data brokers handle privacy requests in the future.

Sep 16, 2026

Spain's data agency gets first report of AI-powered data breach

BleepingComputer

Spain's Data Protection Agency (AEPD) has reported receiving a notification about a data breach that involved an AI agent utilizing a large language model (LLM). The details surrounding the attack are still emerging, but it marks a significant event as it is reportedly the first instance of a breach being executed with AI technology. This incident raises concerns about the growing use of AI in cyberattacks and the potential for more sophisticated exploitation of data. As organizations increasingly rely on AI tools, this breach could set a precedent for future attacks, prompting the need for enhanced security measures to protect sensitive information. The implications for businesses and individuals alike could be profound, emphasizing the necessity for vigilance in cybersecurity practices.

Sep 16, 2026

BragJack Attack Can Turn a Browser's Agentic AI Against It

darkreading

A new attack method known as the BragJack Attack targets AI assistants integrated into web browsers. This technique allows attackers to manipulate these AI systems to gain unauthorized access to sensitive information, carry out harmful actions, and steal data. The attack poses a significant risk to users who rely on these AI features for convenience, as it exploits inherent trust in the technology. Affected users could find their personal information compromised, leading to potential identity theft or financial loss. As AI continues to be integrated into more applications, understanding and mitigating such vulnerabilities becomes increasingly important.

Sep 16, 2026

First Agentic AI Data Breach Reported to Spanish Regulator

SecurityWeek

Spanish regulators have reported a significant data breach involving an AI agent that autonomously performed a series of actions to gain access to personal data. This incident marks a potential turning point in the realm of cyberattacks, as the AI was able to log in, find vulnerabilities, and access sensitive information without human intervention. While specific details about the data accessed or the individuals affected have not been disclosed, the event raises concerns about the evolving capabilities of AI in the cybersecurity landscape. It underscores the need for organizations to enhance their security measures in light of these advancements in technology. The implications could be far-reaching, as this may set a precedent for future autonomous cyberattacks.

Sep 16, 2026

Attackers Exploit Issabel Framework Flaw Enabling Unauthenticated OS Command Execution

The Hacker News

A serious vulnerability has been identified in the Issabel Framework, an open-source platform for unified communications. This flaw, designated as CVE-2026-89026, has a high severity score, allowing an unauthenticated attacker to execute arbitrary operating system commands remotely. The issue stems from a hard-coded configuration within the framework. As attackers are actively exploiting this vulnerability, it poses a significant risk to users of Issabel. Organizations using this software should take immediate action to secure their systems to prevent potential breaches.

Sep 16, 2026