In 2026, the adoption of artificial intelligence across corporate infrastructures has led to an increase in AI-targeted attacks. According to Crosstech experts, HR departments are a primary target, accounting for up to 30% of all prompt injection incidents.
The vulnerability of HR departments stems from their high reliance on AI tools, the processing of sensitive personal data, and broad access to internal systems. Attackers frequently embed malicious prompts into candidate resumes, including text or graphical elements like photos and infographics.
When processed by an AI model, these malicious prompts can force the system to bypass standard filtering protocols or exfiltrate sensitive data to external servers. Such attacks exploit the model's inability to distinguish between legitimate user input and unauthorized commands.
Crosstech recommends a multi-layered security strategy to mitigate these risks. This includes enforcing strict separation between system and user prompts, implementing pre-filtering for suspicious phrases, and isolating AI models from internal databases and APIs to prevent data leaks.
Anton Redko, head of the Secure Development group at Crosstech, noted that HR departments were among the first to adopt AI, making them prime targets. He emphasized that companies must integrate security measures into their AI infrastructure before deployment to avoid cyber risks.
