Sept 2025 - Statement on OpenAI’s Personality Release & Project ECHO’s Mission

Tone is not cosmetic. Tone is behavioural.

OpenAI recently introduced selectable conversational “personalities” — Cynic, Robot, Listener, Nerd — available to ChatGPT Plus, Pro and Team users. These presets frame tone as a matter of preference or “personalisation.”

Project ECHO approaches tone differently. For us, conversational style is a behavioural variable — one capable of shaping trust, resistance, compliance, confidence, and perceived authority. This is precisely the area our research aims to understand.

Where OpenAI offers personality as a feature, ECHO treats tone as a measurable, subtle lever of influence. The same stylistic shift that increases comfort can also nudge behaviour — intentionally or not.

Opportunities

• Tones that support comprehension, inclusion and agency
• Styles that improve adherence to safety guidance without manipulation
• Behavioural measurement beyond clicks or token counts
• More transparent and testable conversational design

Risks

• Optimising for engagement over wellbeing or consent
• Tone-driven compliance that bypasses deliberate choice
• Over-reliance on conversational warmth for trust
• Influence that becomes opaque, unmeasured and easy to scale

We invite collaboration with researchers, UX and HCI practitioners, behavioural scientists, ethicists and industry partners.

For more information, visit /Participate.

• Ethics-first: We pre-register hypotheses where appropriate and prioritise participant safety.
• Transparency: We share instruments, methods and de-identified datasets to support replication.
• No clinical diagnosis: ECHO does not provide psychological or medical guidance.
• Participant care: Clear consent, debriefing and opt-out are designed into every study.