OpenAI announced a new ChatGPT search update in its July 14, 2026 release notes. Users can now search past chats, projects, images, and documents from one place. The feature is rolling out on web, iOS, and Android and is available globally on all ChatGPT plans.
The change addresses a simple but expensive problem: the information exists, but people do not remember which conversation, project, or file contains it. OpenAI says users can start from the ChatGPT sidebar, filter by content type, and open the matching chat, project, or file directly from the result.
For people who use ChatGPT over long-running projects, this makes the product feel less like a sequence of isolated answers and more like a searchable workspace. A single retrieval surface can reduce the need to paste the same background into a new conversation when notes, drafts, images, and documents are spread across several threads.
The release note confirms the search entry point, content-type filters, and direct opening of results. It does not describe ranking behavior, indexing latency, or the permission details between different content types. Businesses should therefore avoid treating this feature as a complete knowledge-base or document-governance system.
A practical rollout starts with using Projects as work boundaries and adopting consistent names for chats and files. Teams should still keep formal document versions, permissions, and approval records in their existing systems. ChatGPT search can reduce retrieval friction without becoming the system of record.
The larger signal is that AI workflows need recoverable context. Once people can find prior work faster, the next layer—organizing, comparing, rewriting, and delivering that work—can become a reusable process instead of a series of disconnected prompts.
