This past weekend, Anthropic’s Claude edged out ChatGPT as the most-downloaded free app in the U.S. And, by early this week, spiking demand seemingly took the tool offline for a number of hours, on two occasions.

For a handful of hours on Monday and again on Tuesday morning, thousands of Claude users reported errors and extended downtime, highlighting a growing business risk as AI, including free tools like Claude, becomes embedded in American work processes.

In online chatter among users, some acknowledged just how reliant they had become on the tool; for instance, many developers admitted they hadn’t written code themselves in months.

“What struck me wasn’t the outage itself. That happens. It was how quickly it exposed how much I’d already baked Claude into my workflow,” wrote one LinkedIn user. “These tools are incredible and I genuinely use them daily. But I think if you aren’t comfortable working without them, then it inherently comes with some risk.”

See also: Workday bets that ‘agents are the future of AI’

Building AI resilience

Cybersecurity, privacy and risk leader Kelvin Flores likened the Claude incident to the 2024 Crowdstrike outage, which was estimated to have caused tens of billions of dollars in financial damages across industries.

Yet, there is a key difference, Flores wrote on LinkedIn.

“CrowdStrike is a security tool. AI is now embedded in how we think, write, analyze and decide,” says Flores. “The dependency runs deeper, and most organizations haven’t caught up on the resilience side.”

A recent CNBC article explored the inherent risks associated with businesses using technology that is advancing more quickly than they can develop their understanding of it.

Mitchell Amador, CEO of security platform Immunefi, told the outlet that business leaders often have “too much confidence” in AI systems, which he called “insecure by default.”

“And you need to assume you have to build that into your architecture,” he said. “If you don’t, you’re going to get pumped.”

The collision of politics and AI

The Claude outage demonstrates another important dimension of the AI landscape leaders are operating in: the inescapable influence of political divides.

Claude’s user surge came as OpenAI—whose ChatGPT was Apple’s most-downloaded app last year—increasingly comes under public scrutiny for a $200 million contract with the U.S. Department of Defense. More than 1.5 million users have signed an online pledge to boycott ChatGPT.

Meanwhile, President Trump banned federal agencies from using Anthropic’s AI tools last week, related to the firm’s opposition to its tools being used in mass surveillance.

Claude reports more than 300,000 business customers and had recently enhanced its knowledge work and customer support offerings. Last month, the company launched Cowork, which enables companies to connect the tool to existing DocuSign, Gmail, Google Drive and other platforms; Cowork plugins are also designed to support everything from HR to engineering workflows, according to the company.

The post Claude outage highlights HR’s growing AI risks appeared first on HR Executive.