Thinking about switching your AI assistant but dreading the cold start? Google just made that leap easier. Gemini now lets users import “memories” and chat histories from rival services like ChatGPT and Claude, so it can instantly personalize responses based on your prior preferences and instructions.
What’s new, and why it matters
According to Google’s guidance, the import works in two parts: first, you ask your current assistant to summarize what it “knows” about you – covering demographics, interests, relationships, dated events, and standing instructions – and paste that summary into Gemini’s new Memory Import tool. Second, you can upload exported chat archives so past conversations appear inside Gemini. The feature is available to free and paid personal accounts, but notably not for work, school, or supervised accounts and not in the UK, Switzerland, or the European Economic Area.
For businesses, this reduces switching costs and onboarding friction. The immediate upside: teams can test-drive Gemini without sacrificing the personalization they’ve already built elsewhere. The obvious risk: moving personal context and chats between providers compounds data governance challenges and legal exposure if sensitive content rides along.
A fast shift toward interoperability
Gemini’s move follows similar portability steps by Anthropic’s Claude, signaling a broader trend: AI platforms are starting to compete less on lock-in and more on play-nice integration. That’s visible elsewhere too. Apple is reportedly building iOS features that let users choose between multiple chatbots via Siri extensions – including ChatGPT, Claude, and Gemini – and has secured broad access to Google’s models, even the option to run distilled versions on-device for performance and privacy gains. If realized, that OS-level routing would make model-switching a tap away, raising the bar for every assistant’s utility and reliability.
On the desktop, the demand for multi-model workflows is already here. ZDNET highlights Noi, a free app that runs ChatGPT, Claude, Gemini, Perplexity, and more in side-by-side windows with session isolation and local-first storage. Together, these signals point to a market that favors portability and orchestration over single-vendor lock-in.
Privacy and compliance: convenience meets accountability
Portability doesn’t erase accountability – and recent events prove it. In a U.S. case reported by Heise, Apple provided law enforcement with the identity and associated iCloud email of a user who had sent a threat while relying on the paid “Hide My Email” feature. That service is designed to reduce spam and keep your primary address private, but it isn’t an anonymization tool. The takeaway for enterprises: “private” features still leave an auditable trail and can be compelled by lawful process. Moving chat logs and personal context between AI vendors increases both the surface area and the paper trail.
There’s also a regulatory dimension to why Gemini’s import isn’t available across Europe. While Google didn’t specify the reason, the region’s stringent rules and enforcement on personal data processing likely raise the bar. For compliance teams, this is a reminder to map feature availability and data flows jurisdiction-by-jurisdiction before rolling out assistant portability at scale.
Policy risk is now a vendor risk
Another underappreciated factor in assistant strategy: government policy can change vendor viability overnight. A federal judge just ordered the U.S. government to rescind its “supply chain risk” label on Anthropic and stop agencies from cutting ties, after a dispute over how its models could be used. Regardless of where one stands on the case, it’s a live example of policy risk affecting AI supplier relationships. Portability features like Gemini’s help enterprises hedge by making multi-model strategies operationally feasible.
How leaders should respond now
- Run a data minimization pass before importing. Strip PII and sensitive client or code references from context summaries and chat logs.
- Update policies: specify which assistants are approved for imported context and where chat archives can be stored.
- Favor vendor-neutral layers. Use OS-level routing (if Apple’s approach ships) or desktop tools like Noi to compare models against the same prompts and data.
- Plan for geography. Document which portability features are available where you operate and build fallbacks for restricted regions.
The bottom line: Gemini’s “memory import” is less about winning a single user and more about accelerating a market shift. As assistants compete on openness and orchestration, professionals will gain leverage – provided they bring rigorous data hygiene and compliance discipline to the party.
Updated 2026-03-30 14:11 EDT: No updates were made to the article as the new source (ID 24483) did not provide significant newsworthy information that would enhance clarity, relevance, or news value beyond what was already covered. The source’s key facts were either redundant or lacked depth, and its news value score of 78 was lower than the target threshold for meaningful addition.