# AI Signals Psychological Safety > AI-generated concerns reveal hidden team tensions without attribution. - Title: AI Signals Psychological Safety · Harvard Business Review - Summary: AI-generated concerns reveal hidden team tensions without attribution. Leaders using AI to surface concerns without naming them can detect psychological… - Keywords: psychological safety, leadership, team dynamics, organizational culture, workplace communication, ai tools, team concerns, business, management, Signals, Psychological, Safety - Source: Harvard Business Review — https://hbr.org/2026/09/what-leaders-need-to-know-about-ai-and-psychological-safety - Published: 2026-09-17T12:25:23+00:00 - Read time: 2 min - Topics: ai, psychological safety, leadership, team dynamics, organizational culture, workplace communication, ai tools, team concerns, business, management ## Safety by proxy Marcus called his team to present a regional expansion plan two days before an anonymous email arrived with a document listing concerns about the expansion. > Someone had asked an AI tool to make the strongest case for the expansion and the strongest case against it. ## AI as anonymous survey The anonymous email contained three pages of concerns about market saturation, customer base size, and unsecured partnerships that were previously unspoken. > The concerns were not necessarily new. What was new was that they were now in writing, circulating, without a name attached. ## Safety by proxy definition Safety by proxy describes using AI to surface concerns when people feel unable to raise them directly in person. > The term describes a substitution, not a solution. ## AI’s perceived neutrality AI outputs arrive with greater status and legitimacy than anonymous surveys, making them a powerful signal of psychological safety gaps. > Safety by proxy describes the specific phenomenon that occurs when people use AI to surface a genuine, substantive concern when they do not feel safe enough to raise it directly. ## Why it matters Research shows team psychological safety directly impacts how teams learn, adapt, and perform under pressure. > One of us (Amy) conducted research on team psychological safety, which is a shared belief that the team is safe for interpersonal risk-taking in service of team goals. ## Leadership’s role Leaders must model fallibility and respond non-defensively to dissent to build psychological safety. > While recognizing that speaking up looks different for different people, and is shaped by personality, communication style, and comfort with direct challenge. ## Four patterns AI’s perceived neutrality interacts with psychological safety in four distinct ways: workaround, equalizer, testing space, and mirror. > If safety is absent, AI can serve as a workaround. If safety is unequal, AI can work as an equalizer. If safety is developing, AI can serve as a testing space. Because even the best leaders are separated by authority, AI can be used as a mirror. ## Workaround pattern People using AI to surface concerns without owning them when speaking up carries personal cost. > The concern enters the room. The person stays safe. AI takes the attribution, at least, for now. ## Equalizer pattern AI helps underrepresented team members bypass communication barriers to voice concerns without attribution. > An underrepresented team member can route a concern through AI and have it arrive without a name attached. ## Testing space pattern Teams use AI to test arguments before raising them directly, signaling growing psychological safety. > Whether this should be seen as accelerating psychological safety or as a substitution remains an open question. ## Mirror pattern AI reveals hidden concerns by bypassing power dynamics that suppress honest input. > The higher a leader’s authority, the more people modulate what they say to reflect what they believe the leader wants to hear. ## 3M case study 3M uses AI to help scientists challenge marketing assertions without direct confrontation. > Established scientists, those with deep technical expertise and years of credibility in their domain, are among the most enthusiastic users of AI-assisted analysis when the conversation moves outside that domain. ## AI stress-testing Leaders should run AI stress-tests before decisions to identify concerns and build psychological safety. > What are the strongest arguments against this direction? What risks haven’t been addressed? What assumptions haven’t been tested? ## Leadership response Leaders must respond to AI-surfaced concerns with curiosity, not defensiveness, to build psychological safety. > Interpret AI-surfaced concerns as a signal, not a threat. ## What happens next Leaders who build psychological safety now can create durable cultural change as AI’s perceived neutrality erodes. > Leaders who build genuine psychological safety now, while AI is still perceived as neutral, can create something durable. ## Key takeaway Leaders using AI to surface concerns without naming them can detect psychological safety gaps before they become critical.