# AI & PS: Research Pulse

*Jade Garratt · Voice & Silence, Interpersonal Practice, Organisational Design*

Jade Garratt reports our research pulse on AI and psychological safety, an open survey run from 25 March to 6 October 2026. Of 110 responses, 96 use AI at work, and 90 of them said how it has affected how psychologically safe they feel with colleagues: 39 saw no change, 31 felt safer (25 somewhat, 6 much) and 20 felt less safe (15 somewhat, 5 much). We had expected AI to sometimes make speaking up a little easier for individuals, especially neurodivergent people, and, in organisations, that it might pull the other way, leaving people feeling more precarious, with less room to get things wrong. Those who felt safer mostly described using AI to prepare for interactions with colleagues: for example, simulating meetings, asking for lines to take, practising emails, stress-testing ideas. Only one tied this directly to neurodivergence. Jade reads it as change in the person, not the team: rehearsal helps people feel more prepared, though the room itself has not changed. One respondent worried that rehearsal could turn into replacement; no one described doing so.

Very few who felt less safe said they were frightened of AI itself. What came through instead was a sense that their own value had become harder to see; others wrote of more pressure and less time to think. Where AI seems to make gaps in knowledge less forgivable, Jade argues, it gets harder to admit what you do not know, and one comment warned of more isolation, more team of one than before. A handful keep their AI use from colleagues, perhaps because in some areas it is considered lazy or even cheating, two of them in higher education. Where AI arrives in a team that already talks openly, and people are given time and space to explore it, Jade concludes, they can talk openly about AI too. Nobody described what Jayshree Seth and Amy Edmondson, writing in HBR, call safety by proxy, using AI to surface a concern they do not feel safe enough to raise directly, but a group that needs a proxy before it can disagree is one where disagreeing directly still feels risky. One reading is that AI amplifies whatever a team already does with uncertainty. The suggestions for managers: go first about your own use; make it ordinary to say you do not know how to do something yet; agree explicitly how the team uses AI; notice anyone who has become a team of one; and welcome rehearsal, encouraging people to make sure the real conversation still happens. It is a small, self-selected sample, likelier to draw people who already care about psychological safety, so a snapshot of experiences rather than a measure of how common they are.

- **This page:** https://explore.psychsafety.com/n/ai-and-psychological-safety/
- **Read the full article on psychsafety.com:** https://psychsafety.com/ai-and-psychological-safety/
- **Interactive map:** https://explore.psychsafety.com/?node=ai-research-pulse

## Neighbours (7)

- [Ambiguity & Predictability](https://explore.psychsafety.com/n/ambiguity-predictability-and-psychological-safety.md)
- [Job Security & PS](https://explore.psychsafety.com/n/job-security-and-psychological-safety.md)
- [PS & Artificial Intelligence](https://explore.psychsafety.com/n/psychological-safety-90-artificial-intelligence.md)
- [PS & Neurodiversity](https://explore.psychsafety.com/n/psychological-safety-and-neurodiversity.md)
- [PS in Education](https://explore.psychsafety.com/n/psychological-safety-in-education-academia-and-teaching.md)
- [Team Charters](https://explore.psychsafety.com/n/psychological-safety-80-team-charters.md)
- [PS with AI / Machines](https://explore.psychsafety.com/n/can-i-feel-psychologically-safe-when-interacting-with-a-machine.md)