Survey: AI Ethics Teams Produced 18,000 Pages of Q2 Guidance; None of It Was Read

SAN FRANCISCO — A survey released Tuesday found that the United States' estimated 4,200 AI ethics professionals delivered a combined 18,000 pages of guidance documentation, 340 responsible-deployment frameworks, and 61 bias-mitigation checklists in the second quarter of 2026, none of which were opened by anyone with budget authority.

The study, conducted by the AI Governance Institute, found that 94 percent of ethics frameworks were distributed via Confluence links that had not been clicked by a non-ethics employee. The average AI ethics hire attended 0.3 product roadmap meetings per quarter, compared with 14.7 for any person with the word "growth" in their title.

AI ethics teams were most frequently consulted in two scenarios: after a regulatory fine had already been issued, and when a journalist had filed a request for comment and someone needed to confirm the company took the issue seriously.

Respondents were asked to rate their organizational influence on a scale of one to ten. The median response was a shrug GIF. The survey instrument was not designed to capture GIFs.

"I have a seat at the table," said one senior ethics lead who asked not to be named because she was not authorized to give on-record statements despite her title being Chief Responsible AI Communication Officer. "It's the table in the hallway outside the conference room, but still."

The survey noted that headcount in AI ethics roles grew 34 percent in 2025, a trend researchers attributed to companies needing someone to write the FAQ for the press release about the thing the AI did.

"In Q3, I expect to be consulted on exactly two items," she added. "The announcement about the AI that got us sued, and the job posting for the AI that's replacing the team that was supposed to implement my framework."