OpenAI has broadened its ChatGPT Deep Research feature, enabling users to connect Microsoft OneDrive and SharePoint accounts. This beta integration allows ChatGPT Plus, Pro, and Team subscribers to analyze their private files within these Microsoft cloud services, marking a significant step in extending AI analysis to personal and enterprise data. The capability aims to boost productivity by allowing users to generate structured, cited reports from their own information.
This allows ChatGPT to access live data from selected OneDrive or SharePoint folders, transforming how users might interact with their internal documents. It follows OpenAI’s recent strategy of adding connectors to external platforms, such as the GitHub integration launched in early May 2025. The new feature is accessible via the ChatGPT composer or settings, as detailed by the OpenAI Help Center.
How The Integration Works And User Access
When a user poses a question, ChatGPT’s Deep Research, now linked to Microsoft services, generates search queries to find relevant information in connected document libraries. These queries are sent to Microsoft, OpenAI explains. The AI then analyzes this data to compile a detailed, cited report, a process that can take five to thirty minutes, differing from ChatGPT’s usual instant answers.
The OneDrive and SharePoint integration is available globally to Team users and is gradually rolling out to Plus and Pro users, according to OpenAI’s release notes. However, it excludes users in the European Economic Area, Switzerland, and the UK. Thurrott.com also clarified that the connector currently supports only work/school OneDrive accounts, not consumer versions. Access for Enterprise and Edu users is expected later.
Strategic Expansion In A Competitive AI Landscape
The introduction of OneDrive and SharePoint connectors continues OpenAI’s push to embed its AI more deeply into user workflows. The earlier GitHub connector was highlighted by OpenAI’s Head of Business Products, Nate Gonzalez, as their “first connector,” responding to user demand for linking ChatGPT to internal data. He noted that users found the Deep Research agent valuable and wanted it connected to their internal sources.
This expansion occurs as the AI research assistant field grows more competitive. Google, for example, made its Gemini Deep Research feature free in March. Other platforms like Perplexity AI, xAI’s Grok 3, and Anthropic’s Claude, along with Microsoft’s Copilot, also offer similar AI research capabilities. By integrating with widely used enterprise tools, OpenAI aims to offer a distinct advantage, particularly for those within the Microsoft ecosystem.
Functionality And Important Considerations
Alongside these connector developments, OpenAI also recently enabled users to export Deep Research reports as professionally formatted PDF files, complete with tables, images, and linked citations. This PDF export feature is available for Plus, Team, and Pro users, with Enterprise and Edu support coming soon.
While the new integration offers powerful capabilities, it’s important to note it’s a beta feature. OpenAI has previously acknowledged that citation accuracy in Deep Research is an area of ongoing refinement.
Furthermore, the underlying o-series models (o3 and the o4-mini) have faced some scrutiny. OpenAI’s own system card and external analyses indicated potential for higher hallucination rates compared to earlier models. OpenAI stated that “more research is needed” regarding this aspect. These considerations are vital as the AI accesses potentially sensitive internal documents, emphasizing the need to verify generated reports, especially during this beta phase. Users should also be aware that connected apps are third-party services with their own terms, as OpenAI clarifies.
