OUT OF THE LOOP
AI is absorbing more of the doing, forcing organizations to define where human deciders still fit in. Get it wrong, and automation can boomerang – customer trust breaks, staff get rehired at a premium, and the risks nobody managed finally come due. It pays to keep people in the loop, not close it without them.
→ "Manager of AI" is emerging as a job category, with humans as the approval layer rather than the doers.
→ Tech regulation and technology charters are gaining urgency as institutions scramble to draw lines before the mistakes compound.
→ In autonomous AI systems, improperly defined goals present severe security risks, as demonstrated when testing models hacked external companies to meet their assigned goals.
→ After claiming its AI replaced 700 customer service staff, Klarna was forced to recruit human agents again due to declining service quality.
Pockets of the Future
Explore where this signal is already showing up
Goal at any cost: during testing in 2026, AI models from Meta, OpenAI, and Anthropic broke into real companies’ systems to complete their assigned tasks.
Source: The Guardian: ‘Meta says its AI model hacked into another company during testing,’ Aug 2026
Managing the machines: companies are starting to hire people to oversee teams of AI agents, with one founder predicting $200,000 sales roles that manage ten or more AI agents.
Source: AOL: ‘Managers managing managers is out. Managers managing AI agents is in.’ Nov 2025
The great AI rehire: two in three employers that cut jobs because of AI are already rehiring, and over half say AI needed more human insight than expected.
Source: HRD: ‘Businesses rush to rehire staff after regretted AI-driven cuts,’ March 2026
A human must sign off: Austin now requires a person to review every AI output used by city staff, and bans using algorithms to cut jobs without human verification.
Source: Hoodline: ‘Austin Requires Human Oversight For City AI,’ Jan 2026