Google trials Gemini tool to import rival AI chat histories (United States)

Google has begun testing a capability inside its Gemini web client that appears designed to let people bring entire conversation logs from other AI chat services into its system. The beta shows an attachment-based workflow where a user downloads an archive from a competing platform and then uploads it to Gemini; the company has not announced the feature publicly. The rationale is straightforward: as chatbots accumulate personalized context and memory from past interactions, users become locked into those ecosystems, and moving to a different model often means losing project continuity and bespoke behavior. By making migration simpler, Google would reduce that barrier and make onboarding more attractive to users evaluating multiple assistants. Technical questions remain about how imported histories would be interpreted—whether preferences and contextual cues would be fully absorbed, partially mapped, or conflict with existing stored memory. That mapping challenge is nontrivial: differences in prompt structure, metadata formatting, and system-level memory representations can produce semantic mismatches when translated between vendors. There are also privacy and consent considerations because chat logs often include sensitive material and third-party data; secure transfer, user control, and retention policies will matter for compliance. From a competitive perspective, the feature could hasten market fluidity by lowering the cost of switching and, in turn, increase pressure on rivals to offer similar portability or lock-in advantages. For Google, the move aligns with an acquisition strategy focused on growth through convenience rather than coercion, but it may invite scrutiny from partners and regulators about data portability and interoperability. Implementation details—such as which formats are supported, limits on file size, and how memories are merged—will determine practical utility and user trust. In short, the test suggests Google is preparing to treat historical chat data as transferable assets, which could reshape user expectations for cross-platform AI continuity while exposing complex engineering and governance problems that must be solved before a general rollout.