A board director approved a customer-service layoff in 2023 citing AI capability. Eighteen months later, the company is hiring those roles back. Offshore. Lower pay. Without a press release.

This is not a one-off. Gartner forecast on February 3rd, 2026 that 50% of companies cutting customer-service staff citing AI will rehire by 2027. Forrester’s Predictions 2026 put it more directly: when CEOs announcing AI-driven layoffs were asked whether mature AI was actually in place at the time of the cuts, the answer was no nine times out of ten.

Here is what changed in the last 24 months — and what every board needs to start asking before approving the next AI-driven workforce action.

What’s actually working

The productivity gains from AI are real. Stanford HAI’s 2026 AI Index reported $172 billion in annual consumer value generated by generative AI tools in early 2026. McKinsey reports high performers using AI are 2.8 times more likely to have redesigned workflows around it. Engineering teams using AI code assistance consistently report 20 to 30% velocity improvements on bounded tasks. Customer-service agents with AI co-pilots resolve calls faster and escalate less often than agents working alone.

And the trajectory is real. The 24% completion ceiling on Carnegie Mellon’s TheAgentCompany benchmark in December 2024 became 30.3% within months. The capability gap between frontier models and the prior generation closes every six to twelve months. The companies that have bet against AI capability improving have lost that bet for fifteen years running.

This is the promise. Every board should be looking at it seriously.

Where the ceiling shows up

The same capability data also shows a hard ceiling, and that is where the quiet rehire comes from. Twenty-four percent task completion is not the threshold for wholesale replacement in judgment-bearing work. MIT NANDA found 95% of enterprise GenAI pilots produced no measurable P&L impact despite $30 to $40 billion spent. McKinsey’s State of AI 2025 reported that only 39% of organizations using AI can point to EBIT impact, and only 23% are scaling AI agents beyond pilot.

The Klarna case made this concrete. By May 2025, CEO Sebastian Siemiatkowski admitted in a Bloomberg interview that the company’s chatbot-first strategy had produced lower quality, and Klarna began rehiring humans. Commonwealth Bank of Australia reversed 45 customer-service layoffs in August 2025 under union pressure, with the bank’s own statement admitting the initial assessment “did not adequately consider all relevant business considerations.”

Two companies. Two jurisdictions. Same failure mode.

The governance exposure compounds. Delaware Caremark doctrine already covers AI workforce decisions where the technology is core to operations. EU AI Act Article 4 enforcement begins August 2nd, 2026. The SEC has already settled AI-washing charges twice (Delphia and Global Predictions, March 2024) and the architecture is built for a workforce case. Deloitte’s 2025 study of 695 board members found 66% describe their boards as having limited to no AI knowledge.

The sequence that survives Delaware

The Promise & Risk needle leans toward Promise, but only for organizations that treat the four stages as discrete governance gates, not slide titles.

Augment humans first in judgment-bearing workflows. Validate against two bars at once: cost-acceptable use-case fit and capability trajectory (re-baseline every six months). Consolidate with workflow redesign, not bolt-on automation. Redeploy honestly, distinguishing genuine higher-leverage work from severance with extra steps. Disclosure language has to be consistent across the press release, the board minutes, and the all-hands. Inconsistency is what attracts SEC attention.

Cutting first and validating later is what produces the quiet rehire. The companies treating governance as a precondition for ROI, not a parallel workstream, will be the ones that compound. The ones reading the headlines instead of the data are about to discover that governance debt and technical debt behave the same way. They don’t go away. They get more expensive.

For the full analysis → The longer piece on fredriklindstrom.info walks through the four-stage sequence in detail, with the legal and regulatory exposure mapped against each step, plus the four-question board test before approving any AI-driven workforce action. Read it here: The Quiet Rehire.

Want every Promise & Risk article in your inbox? Subscribe to the Promise & Risk weekly digest →