Klarna has been held up as an example of success with AI teams…. Well… Futurism reports Klarna has faced significant challenges after replacing its customer support team with artificial intelligence. Following a staggering 17 percent default rate on its loans, Klarna’s CEO Sebastian Siemiatkowski initially boasted about the efficiency of AI in handling customer service tasks, even claiming it replaced the work of 700 full-time agents. However, by May 2025, the company began rehiring human staff, struggling to restore its customer support department, and has resorted to reassigning engineers and marketers to handle calls due to the AI’s shortcomings. As Klarna approaches its anticipated stock market debut, which is targeting a valuation of $14 billion, it remains uncertain if it can successfully transition its technical staff to customer support roles.
A recent study reveals that while corporate artificial intelligence initiatives often fail to yield measurable returns, employees are increasingly adopting consumer-grade AI tools on their own, creating a “shadow AI economy.” According to Gallup, 40 percent of U.S. workers now use AI at work, nearly double the rate from two years ago, driven by the need to manage growing workloads with greater efficiency. Experts suggest that this grassroots adoption may lead to productivity gains, but also introduces risks related to compliance, security, and operational reliability.
Why do we care?
Klarna thought it was clever—fired 700 support staff, bragged AI could do it better. Fast forward, customers are furious, loan defaults spike, and now? They’re dragging engineers and marketers onto the phones. That’s not efficiency—that’s chaos.
Meanwhile, Gallup says 40% of U.S. workers are using AI at their jobs already. That’s double from two years ago. And most of it? It’s “shadow AI”—people just pulling in consumer tools because their bosses haven’t given them anything official. So what’s the takeaway? You can’t ban it, people will use it anyway. But you also can’t rip and replace humans with bots and hope it works. The smart play—especially for us in IT services—is to help companies put guardrails around this. Policies, governance, security. That’s the sweet spot. Manage AI adoption before it manages you.