TL;DR
According to a Wired report, OpenAI and data firm Handshake AI have asked third-party contractors to upload real deliverables from past or current jobs and describe the tasks they performed. The request includes guidance to remove proprietary data and personal information, but a lawyer warned the practice could create significant legal and confidentiality risks.
What happened
Wired reports that OpenAI and training-data company Handshake AI have been asking third-party contractors to provide examples of real work they produced in past or current roles. A company presentation reportedly instructs contractors to describe tasks they completed and to upload concrete outputs — not summaries — such as Word documents, PDFs, PowerPoint slides, Excel files, images, or repositories. Contractors are told to remove proprietary information and personally identifiable data before uploading and are directed to a ChatGPT tool labeled 'Superstar Scrubbing' to assist with redaction. The request appears related to a broader industry trend of hiring contractors to create higher-quality training data aimed at automating more white-collar tasks. An intellectual property lawyer cautioned that relying on contractors to judge confidentiality could expose a company to legal risk. OpenAI declined to comment, according to the report.
Why it matters
- Using real, on-the-job outputs as training inputs raises ownership and intellectual property questions about who controls the underlying work.
- Relying on contractors to identify and remove confidential or personal data shifts significant trust and legal responsibility onto external workers.
- High-quality, task-specific examples could accelerate automation of professional services if used to train models, changing labor dynamics.
- The approach reflects an industry pattern of sourcing curated human-created data for model development, which may prompt scrutiny from clients and regulators.
Key facts
- Wired reported that OpenAI and Handshake AI have asked contractors to upload actual work they produced in past or current jobs.
- The company presentation reportedly asks contractors to describe tasks they performed and to provide the concrete output files, not summaries.
- Accepted example file types include Word documents, PDFs, PowerPoint decks, Excel sheets, images, and code repositories.
- Contractors are instructed to remove proprietary information and personally identifiable information before uploading.
- OpenAI pointed contractors to a ChatGPT 'Superstar Scrubbing' tool to help delete sensitive data.
- An intellectual property lawyer told Wired that the practice could expose an AI lab to significant legal risk by putting confidentiality judgments on contractors.
- OpenAI declined to comment for the report.
- The practice is described as part of a broader industry effort to hire contractors to generate higher-quality training data for automating white-collar work.
What to watch next
- Whether other AI companies publicly adopt similar programs to collect contractor-supplied work — the report links this to a broader industry trend.
- If uploaded contractor files are actually incorporated into model training datasets or used in production systems (not confirmed in the source).
- Any legal actions, complaints, or regulatory inquiries arising from the use of contractor-provided work (not confirmed in the source).
- Whether OpenAI updates its guidance, tooling, or oversight processes for contractors after the report (not confirmed in the source).
Quick glossary
- Contractor: An external worker or third party engaged to perform specific tasks or projects for a company, often on a temporary or freelance basis.
- Training data: Examples and datasets used to teach a machine-learning model to recognize patterns and produce outputs.
- Proprietary information: Business-owned information that gives a company a competitive edge and is often protected by confidentiality agreements.
- Personally identifiable information (PII): Data that can be used to uniquely identify or contact an individual, such as names, addresses, or social security numbers.
- Data scrubbing: The process of removing or anonymizing sensitive or confidential information from files before they are shared or used.
Reader FAQ
Are contractors being asked to upload real work they did at prior jobs?
Wired reports that OpenAI and Handshake AI asked contractors to provide concrete outputs from past or current jobs and describe the tasks performed.
Did OpenAI respond to the report?
According to the article, an OpenAI spokesperson declined to comment.
What types of files were contractors asked to upload?
The reported examples include Word docs, PDFs, PowerPoint files, Excel spreadsheets, images, and code repositories.
Does the report confirm these uploads were used to train models?
Not confirmed in the source.
Are there legal concerns with this approach?
An intellectual property lawyer told Wired that the practice could expose a company to substantial legal risk by depending on contractors to judge confidentiality.
IN BRIEF Posted: 1:18 PM PST · January 10, 2026 IMAGE CREDITS: JAKUB PORZYCKI/NURPHOTO / GETTY IMAGES Anthony Ha OpenAI is reportedly asking contractors to upload real work from past…
Sources
- OpenAI is reportedly asking contractors to upload real work from past jobs
- OpenAI Is Asking Contractors to Upload Work From Past …
- To prepare AI agents for office work, the company is asking …
- OpenAI Is Paying Ex-Investment Bankers to Train Its AI
Related posts
- FCC Approves SpaceX Plan to Launch 7,500 More Gen2 Starlink Satellites
- How the High School You Attend Reshapes Your Chances for UC Admission
- Timeline: Minnesota’s Standoff With Federal Government Over Fraud and Immigration