OpenAI Trained Its AI on Workplace Data: Should Employees Be Worried?

OpenAI is reportedly asking third-party contractors to upload real work they have completed in past and current jobs as part of an effort to improve the quality of training data used for its AI models. 

According to reporting from Wired, OpenAI is working with a training data company called Handshake AI to collect authentic workplace materials created by contractors, rather than relying solely on synthetic or hypothetical examples. This initiative appears to reflect a broader trend across the AI industry, where companies are increasingly turning to human contractors to generate realistic data in hopes of building models capable of automating more complex white-collar tasks (Wired, qtd. in TechCrunch).

As part of this process, OpenAI reportedly asks contractors to describe tasks they performed at previous jobs and upload actual examples of the work they produced. These materials can include Word documents, PDFs, PowerPoint presentations, Excel spreadsheets, images, or even code repositories. Importantly, the company emphasizes that these submissions must be real outputs rather than summaries or recreated examples. The goal is to expose AI systems to genuine, real-world work products so they can better understand how office tasks are completed in practice, not just in theory (Yahoo Tech).

OpenAI has reportedly put safeguards in place to address privacy and confidentiality concerns. Contractors are instructed to remove any proprietary or personally identifiable information before uploading files. To assist with this, OpenAI provides access to a tool referred to as the ChatGPT “Superstar Scrubbing” tool, which is designed to help contractors identify and remove sensitive data. Despite these precautions, the responsibility largely falls on individual contractors to determine what information is safe to share, creating potential areas of risk if mistakes are made during the scrubbing process (Digit).

Legal experts have raised concerns about this approach, particularly around intellectual property and confidentiality. Evan Brown, an intellectual property lawyer, told Wired that any AI lab adopting this method is “putting itself at great risk” because it relies heavily on contractors to judge what is confidential and what is not. Even with tools and instructions in place, there is no guarantee that sensitive or proprietary information will be fully removed, especially when dealing with complex workplace documents. An OpenAI spokesperson declined to comment on the matter when asked about the practice (TechCrunch).

The reported initiative aligns with a larger push by AI companies to develop systems that can handle everyday office work such as drafting emails, building presentations, analyzing spreadsheets, and managing documents. By studying real human-generated work, AI developers hope to train models that can replicate these tasks with greater accuracy and usefulness. Proponents argue that real-world data is essential for creating AI agents that function effectively in professional environments, where context, formatting, and nuance matter (Digit).

At the same time, the practice raises broader ethical questions about data use, consent, and ownership. Even when contractors voluntarily submit their work, concerns remain about whether employers from past jobs are aware their materials could be used to train AI systems. As AI companies continue to seek more realistic training data, debates around privacy, intellectual property, and transparency are likely to intensify, highlighting the tension between technological progress and the protection of human-created work (Yahoo Tech).

This article was written by the Covertly.AI team. Covertly.AI is a secure, anonymous AI chat that protects your privacy. Connect to advanced AI models without tracking, logging, or exposure of your data. Whether you’re an individual who values privacy or a business seeking enterprise-grade data protection, Covertly.AI helps you stay secure and anonymous when using AI. With Covertly.AI, you get seamless access to all popular large language models - without compromising your identity or data privacy.

Try Covertly.AI today for free at www.covertly.ai, or contact us to learn more about custom privacy and security solutions for your business.  

Works Cited

Digit. “OpenAI Asks Contractors to Upload Real Work from Past Jobs to Test AI Agents: Report.” Digit, 2026, www.digit.in/news/general/openai-asks-contractors-to-upload-real-work-from-past-jobs-to-test-ai-agents-report.html.

TechCrunch. “OpenAI Is Reportedly Asking Contractors to Upload Real Work from Past Jobs.” TechCrunch, 10 Jan. 2026, www.techcrunch.com/2026/01/10/openai-is-reportedly-asking-contractors-to-upload-real-work-from-past-jobs/.

Yahoo Tech. “OpenAI Is Reportedly Asking Contractors to Upload Real Work from Past Jobs.” Yahoo Tech, 2026, www.tech.yahoo.com/ai/articles/openai-reportedly-asking-contractors-upload-211829411.html.

Tong, Anna, Jeffrey Dastin, and Krystal Hu. “Exclusive: OpenAI Researchers Warned Board of AI Breakthrough Ahead of CEO Ouster, Sources Say.” Nasdaq, 22 Nov. 2023, https://www.nasdaq.com/articles/exclusive-sam-altmans-ouster-at-openai-was-precipitated-by-letter-to-board-about-ai-breakthrough-sources-tell-reuters.

Knight, Will, et al. “OpenAI Launches ‘Deep Research’ Tool That It Says Can Match Research Analyst.” The Guardian, 3 Feb. 2025, https://www.theguardian.com/technology/2025/feb/03/openai-deep-research-agent-chatgpt-deepseek.

“OpenAI Is Asking Contractors to Upload Work From Past Jobs to Evaluate the Performance of AI Agents.” Wired, 9 Jan. 2026, https://www.wired.com/story/openai-contractor-upload-real-work-documents-ai-agents/.