OpenAI Removed Contractors for Using AI

Contractors were fired for utilizing unauthorized AI tools to complete mandatory human assessment tasks.

Updated on Sept. 25, 2026 in Artificial Intelligence

Isometric editorial illustration of a stack of translucent drafting films on a wooden desk, representing manual data review processes.
OpenAI has terminated multiple contractors supplied by Mercor after discovery that they used unauthorized artificial intelligence tools to complete mandatory human evaluation tasks. AI Illustration. Upload story photo >

Live Poll

Do you believe it is acceptable to use AI tools to complete professional work assignments?

OpenAI has removed multiple contractors from projects after they were discovered using artificial intelligence to complete assessment tasks. These contractors were tasked with manually reviewing user prompts and chatbot responses to ensure high-quality output.

Why it matters

The company relies on human review to refine its systems, and unauthorized AI use threatens the integrity of the data used to train its models. Ensuring that assessments are completed without software assistance remains a core requirement for those evaluating ChatGPT-related material.

Internal guidance explicitly prohibits the use of AI tools, translation software, Grammarly, and AI detection programs for these tasks. Human reviewers are required to perform all quality assessments independently to meet company standards.

The players

OpenAI

This research organization is the creator of ChatGPT and utilizes human contractors to help improve its large language models.

Mercor

Mercor is an AI-training company that provides the contracted workforce tasked with reviewing and auditing AI model outputs.

The details

Mercor, an AI-training firm, supplies the workforce responsible for reviewing ChatGPT-related material. Internal protocols mandate that all human reviewers complete their prompts and response assessments without any form of software assistance.

Timeline

  1. September 25, 2026

The Tech Race

This development highlights the ongoing struggle to maintain high-quality data labeling as the industry relies on RLHF to refine model behavior. By removing contractors, OpenAI demonstrates its focus on preventing synthetic data pollution within its foundational training pipelines.

Users interacting with AI systems should remain aware that model improvements depend on human accuracy and the integrity of review processes. The removal of these contractors signals that the company is actively tightening oversight to prevent automated shortcuts in data quality.

The takeaway

Maintaining the quality of AI models remains a labor-intensive task that defies simple automation. As these tools advance, companies will continue to face challenges in ensuring that their human workforce remains the final arbiter of machine intelligence.

Further reading

For more context on the industry, visit our Artificial Intelligence section.

Source note: This article includes information reported by Scottish Legal News.

Live Poll

Do you believe it is acceptable to use AI tools to complete professional work assignments?