Futures Report · 26Q3 · Q3 2026
Who Answers for the Agents?
Accountable AI and the return of human ownership
Agents are being given job titles and org-chart lines. Accountability, though, has never belonged to the technology — it belongs to a named human.
- Authors
Diana Wu DavidFutures Director
Simon GriceSenior Director of Innovation
- Length
- 19 pages
- Themes
- Trust & Security, The Actualized Enterprise, AI Strategy & Maturity
Summary
- While only one in five organizations has a mature AI governance model, AI agents are getting job titles and org-chart lines. The employee label risks weakening human oversight.
- Accountable AI is the discipline of making humans, through the organization deploying the AI, answerable for agents' outcomes. It has five parts — S.T.E.E.R.: Standards, Transparency, Effects, Enforcement and Responsibility.
- Regulators and courts have long determined accountability sits with a human or an organization, never with the technology itself.
- Organizations that assign clear ownership report higher governance maturity, stronger trust and better returns on their AI investment.
- One named person answers for each use case, the way a director answers for a company, even when they cannot watch every single operation.
- The technological and organizational foundations for accountable AI are buildable today — and more urgently needed than ever.
From the brief
A Formula 1 pit stop takes under three seconds. Twenty people are already waiting: one on each wheel gun, one on each jack. Nobody does two jobs.
Above them, behind a bank of screens, sit the decision-makers, watching the data the car streams back in real time and holding the authority to call it in early or retire it altogether.
Read the full report