ChatGPT rolled out its memory feature to Plus subscribers in February 2024, with broader availability through the rest of the year. The pitch was simple: the model remembers things you tell it across conversations. What some users are noticing now is something subtler - the assistant's personality gradually shifting to match how they actually communicate.
This goes beyond stored facts. ChatGPT's memory (accessible under Settings > Personalization > Manage Memory) holds discrete data points: your job title, your preferred output format, ongoing project names. But communication style is fuzzier than any single stored fact. If you consistently write in short sentences, ask direct questions, or prefer bullet points over paragraphs, the model starts mirroring that without you explicitly requesting it.
What's Actually Happening
OpenAI hasn't published specifics on how personality adaptation works, but the observable behavior matches what you'd expect from a model that weights recent interaction history heavily. The more you use it in a consistent style, the more that style gets reflected back. It's less about explicit memory storage and more about pattern reinforcement across a long conversation history.
This is worth separating from custom instructions, which you write explicitly in settings. Adaptive personality happens passively - easier to miss if you're not paying attention to changes in response tone over weeks.
The Practical Trade-offs
For people who use ChatGPT daily across consistent work - drafting emails, writing code, researching topics - the adaptation is genuinely useful. The assistant stops feeling like a generic tool and starts matching your rhythm. Fewer clarifications about what you actually want, less time reformatting outputs.
The downside is unpredictability. If your work spans very different modes - technical documentation one day, casual copy the next - the personality drift can feel inconsistent. And there's no dashboard showing how your interaction style has shaped the model's behavior over time. You're adapting to something you can't fully see.
This accumulated context is also the strongest argument for staying in one AI tool rather than bouncing between them. ChatGPT doesn't know you if you've been using Claude for three months. The switching cost is higher than it used to be - and it's getting higher the longer you stay.