AI chatbots are helping find more software security flaws

In short: Easy-to-use AI chatbots are making it faster for people to find software security flaws, even as some AI labs discuss slowing down advanced AI work.

What's going on

Security researchers are reporting a rise in newly found software vulnerabilities, which are weaknesses that can let attackers break in. A key reason is that AI chatbots are getting good at scanning code and suggesting where problems might be.

Think of it like having a very fast proofreader for software. It can point out suspicious lines and patterns that a human might miss, or would take hours to review. This can help defenders fix issues sooner, but it can also help criminals find the same weak spots.

The WIRED newsletter Kernel Panic highlights a tension in the AI world. Some big AI labs have been talking about an industry-wide pact to slow down development of the most powerful systems. At the same time, widely available tools are already spreading the ability to look for bugs beyond specialist security teams.

This is not just about one company or one product. The trend is about access. When helpful tools become common and cheap, more people can do the work, for better and for worse.

What to watch

Watch for more pressure on software makers to patch faster, and for more rules about how powerful AI systems are shared. Also watch whether companies invest more in testing tools that look for weaknesses before software ships, like checking a house for unlocked doors before moving in.

Source: Wired