A financial services firm started using Cursor, an AI tool that writes code on its own. Output jumped from 25,000 lines per month to 250,000. That left a backlog of one million lines needing review, according to Joni Klippert, co-founder and CEO of security startup StackHawk.
“The sheer amount of code being delivered, and the increase in vulnerabilities, is something they can’t keep up with,” she said. The speed also pushed sales, marketing, and support teams harder, adding a lot of stress.
Why Code Overload Hit So Fast
AI coding tools from companies like Anthropic, OpenAI, and Cursor exploded last year. Now anyone can spin up software ideas in hours, not weeks. Tech workers call it a superpower for brainstorming, but reviewing the flood of code for bugs and risks falls on too few engineers.
A Google survey showed 90% of developers now use AI at work, with 71% relying on it for actual coding. Open source projects on platforms like GitHub got swamped with AI-generated contributions, some spammy or flawed.
Companies Struggle to Keep Up
Senior engineers who spot errors fast are suddenly in high demand. Tech firms including Pinterest, Block, and Atlassian cut jobs, pointing to AI efficiencies. Meta’s CTO Andrew Bosworth noted projects once needing hundreds of engineers now take tens, with work dropping from months to days.
“The blessing and the curse is that now everyone inside your company becomes a coder,” said Michele Catasta of Replit.
New headaches popped up too: unclear ownership of AI-generated bugs, security risks from downloading company code to laptops, and overwhelmed review processes. Some startups now require human review for all code to avoid hidden breaks.
Fixes on the Horizon
Tools are adapting. Cursor bought Graphite to build better code-review bots. Anthropic and OpenAI released AI agents that scan for errors. The old way of writing a few dozen careful lines a day is gone, replaced by agents that generate full programs with minimal guidance.
