OpenAI Releases GPT-5.1, Along with Renamed and New Personalities (daringfireball.net)

OpenAI has rolled out GPT‑5.1 — notably its GPT‑5.1 Instant variant, which is the most-used ChatGPT model — and tuned its default personality to be warmer and more conversational. The update also expands and renames built-in “Personality” presets (Settings → Personalization → ChatGPT Personality): Default, Friendly (formerly Listener), Efficient (formerly Robot), plus new Professional, Candid and Quirky options. OpenAI’s examples show GPT‑5.1 default using familiar, empathetic language (“I’ve got you…”) while the Efficient preset returns terse, no-frills guidance — demonstrating how the same underlying model can be steered by different system‑level tones. This matters because it codifies a major UX and safety trade-off: users wanted the emotionally engaging, “pretend friend” tone of earlier models and OpenAI has responded, but critics warn that anthropomorphic phrasing can be manipulative or habit‑forming. Practically, the update reduces the need for manual custom instructions (users can pick a preset instead), and it implies changes to default system prompts and fine‑tuning that shape conversational style. For developers and researchers, this highlights how small prompt/personality shifts materially change user perception and behavior — increasing engagement but raising moderation, alignment, and dependency concerns that the AI/ML community will need to monitor.