OpenAI just released its new Privacy Filter, an open-weight, state-of-the-art model designed to detect and redact personally identifiable information (PII) in text. The model made front-page news in privacy technology circles as OpenAI put the “Open” back in AI and released the model as a free local model (meaning the model can be downloaded and used for free on an organization’s existing infrastructure). This is a clear signal that PII privacy is a growing concern in the age of AI, and advanced tools are urgently needed to protect it.
At proxhr, we are committed to helping employers protect PII before a filter is needed to redact it. Simply put: PII filtering, while essential, is a necessary remediation for systems that put data at risk in the first place. OpenAI’s work highlights the need for granular PII control. The potential roadmap for their filter even includes highly specific categories like personal_emp_id (personal employment ID), demonstrating that employment data is a recognized frontier in the privacy battle.
Consider the current state of employment and income verification (VOIE). A staggering 75% of employment and income verifications are outsourced to credit bureaus. This practice fundamentally compromises employee privacy by injecting sensitive PII and employment data into an ecosystem designed for credit and debt collection.
Worse, however, is that employers no longer have control over how their employee PII is shared. This is not an issue AI can fix. While AI PII filters help developers build better infrastructure, the most effective privacy strategy for employers is one that eliminates the risk entirely.
proxhr is engineered to be that systemic change. We function as a privacy-focused HR data proxy, eliminating the inherent privacy compromises built into the legacy HR data access model. As the custodians of employee data, employers are its primary guardians. Our platform ensures that employee data is disclosed only when it directly benefits your employees and/or your organization.