Payroll Accountants Risk Data Breaches by Uploading Payslips to Public AI Tools
Translated & summarized from Walla by baba
Israeli payroll accountants increasingly use public AI tools to process payslips, risking serious data breaches and legal violations. A survey found 85% uploaded payslips to AI platforms, exposing sensitive payroll data to foreign servers and potential misuse. This practice breaches Israeli privacy laws, which require strict protection of payroll information. Experts urge specialized training and use of secure AI systems to prevent legal and operational risks.
The story in 6 lines · by baba
- 85% of payroll accountants uploaded payslips to public AI systems in the past year, risking data exposure.
- Use of public AI tools like ChatGPT transfers sensitive payroll data to foreign servers without proper security.
- Uploading payroll data to public AI violates Israeli privacy laws with penalties up to five years imprisonment.
- AI improves payroll processing by consolidating data and reducing errors, but only when used securely.
- Oketz Systems launched a training course to help accountants use AI legally and effectively.
- From 2025, explicit consent will be required for AI processing of personal payroll data under new privacy regulations.
Payroll accountants in Israel are increasingly using unprotected artificial intelligence (AI) tools to consult on professional issues by uploading sensitive payslips, rather than relying on traditional or secure organizational platforms. According to a survey by Oketz Systems, 85% of payroll accountants uploaded payslips to AI systems in the past year, while activity in professional Facebook groups dropped by about 45%. This shift reflects AI’s growing role in streamlining payroll processes, such as consolidating attendance, commissions, vehicle benefits, and expense reimbursements from various sources into a single table for payroll software, reducing errors and workload.
However, many accountants are uploading payslips to public AI platforms like ChatGPT, which poses significant risks. These platforms transfer data to foreign commercial servers and subcontractors, potentially using the information to train future AI models unless explicitly blocked. This exposes organizations to threats including credential theft, data leaks, and phishing attacks targeting payroll details, which can lead to salary theft through bank account manipulation.
This practice violates Israeli privacy laws, specifically Amendment 13 to the Privacy Protection Law, which classifies payroll data as highly sensitive and mandates strict security. Unauthorized disclosure can result in up to five years imprisonment, and there is a legal obligation to secure such data. Additionally, no prior agreement is typically in place with external AI providers, breaching regulatory requirements. From 2025, the Privacy Protection Authority will require explicit consent for personal data processing via AI, effectively banning the input of identifiable payroll data into public AI tools.
Chaim Molcho, co-CEO of Oketz Systems, emphasized that while AI is revolutionizing payroll accounting, legal compliance and specialized training are essential to avoid legal exposure and operational errors. Oketz has launched a course to guide accountants on safely integrating AI. Molcho warned that generic AI tools not tailored to specific office workflows risk producing incorrect outputs, which may only be detected too late, undermining payroll accuracy and security.