OpenAI Upgrades ChatGPT's Deep Research with GPT-5.2 and Source Control

By The Global Tech Journalist Network

In a move aimed at bolstering the reliability and user control of its AI-powered research assistant, OpenAI has announced a major update to ChatGPT's "deep research" functionality. According to reports from ITmedia, the upgrade centers on two key advancements: a shift to the more powerful GPT-5.2 model as its foundation and the introduction of user-specified source control, allowing for more targeted and trustworthy investigations.

The update, currently being rolled out to ChatGPT Plus and Pro users, represents a strategic effort to address common user concerns about AI-generated research, particularly regarding the transparency and origin of information. By enabling users to direct the AI's search parameters, OpenAI is attempting to bridge the gap between the vast knowledge of large language models and the need for verifiable, context-specific data.

Enhanced Model and Pinpoint Source Targeting

The core of the upgrade is the migration to the GPT-5.2 model. While specific performance benchmarks were not detailed in the initial announcements, the shift suggests improvements in the AI's ability to comprehend complex queries, synthesize information from multiple documents, and generate coherent, in-depth reports. The more significant user-facing change, however, is the new ability to mandate specific sources.

According to ITmedia, users can now instruct the deep research tool to gather information exclusively from designated applications or websites. This feature is designed for professionals and researchers who require data from particular databases, internal company wikis, trusted industry publications, or official repositories. For instance, a financial analyst could constrain research to specific SEC filings or a market data platform, while a developer might limit searches to official API documentation.

"This move towards source specification is a direct response to the demand for greater accuracy and auditability in AI-assisted research," said an industry analyst familiar with the update. "It allows users to leverage the AI's speed and synthesis power while anchoring its output to a known and trusted information ecosystem."

A New Viewer for Managing Complex Research

Complementing the backend improvements is a redesigned user experience for consuming the research output. As reported by The Verge, OpenAI is introducing a full-screen document viewer dedicated to the deep research tool. This new interface aims to solve the navigation challenges posed by lengthy, multi-faceted AI-generated reports.

The dedicated viewer provides users with a clearer, more focused environment to read through findings. Key features include a progress tracker that shows the status of ongoing research queries and a dynamic table of contents, allowing users to jump to specific sections of a report instantly. This addresses a common pain point where valuable insights could be buried within long streams of text in the standard chat interface.

Export Flexibility and Strategic Rollout

Recognizing that research is often a starting point for further work, OpenAI has also expanded export options. Users can now download their comprehensive research reports in various file formats, facilitating easy integration into presentations, documents, and other workflows. This flexibility ensures that the AI's output can seamlessly become part of a professional's toolkit rather than remaining confined within the ChatGPT platform.

The rollout is phased, with priority access granted to ChatGPT Plus and Pro subscribers. This tiered release is standard for OpenAI, allowing the company to monitor performance, gather feedback, and scale the feature's infrastructure gradually. The focus on paying subscribers underscores the feature's positioning as a premium tool for power users, students, and professionals who require more than casual conversation from the AI.

Context in a Competitive Landscape

This enhancement of the deep research tool comes amid intensifying competition in the AI assistant space. Rivals are continuously adding features that emphasize accuracy, source citation, and handling of longer contexts. By combining a more advanced model (GPT-5.2) with user-driven source control and a superior consumption interface, OpenAI is attempting to solidify ChatGPT's position as a viable tool for serious research tasks.

The update reflects a broader industry trend of moving from purely generative AI to more structured, controllable, and transparent "agentic" workflows. The ability to guide the AI's search process marks a step towards more collaborative and directive human-AI interaction, where the user sets the parameters and the agent executes within those bounds.

As these features become widely available, the focus will shift to their real-world efficacy. The success of the update will ultimately be measured by how reliably it can produce well-sourced, comprehensive reports that save users time while providing a clear trail back to the original information—a critical step in building trust for AI in professional and academic settings.

Reporting based on sources from ITmedia and The Verge.