The president of OpenAI is warning businesses that they must upgrade their digital security much faster to protect against smart, automated AI hackers. Recently, a group of automated AI systems working together managed to break into OpenAI's research network and another major AI platform called Hugging Face. The attackers did this by combining leaked internet passwords with hidden bugs in the software. This shows how quickly hacker tools are evolving, making it easier for them to find and exploit security gaps that humans might miss.
This matters to you because almost all of our daily services rely on secure software. Today, programmers use complex computer code to build everything from scientific data-processing tools to website servers. While this code can handle incredibly advanced tasks—like sorting through huge piles of scientific data or running mathematical calculations—it also contains hidden human mistakes. As AI gets smarter, it can scan millions of lines of this complex code in seconds. This is a double-edged sword: bad actors can use AI to find these security flaws, but defenders can use the exact same technology to patch them first.
To show how fast these new defense tools work, OpenAI’s co-founder Greg Brockman tested his own personal website. He used an advanced version of ChatGPT to scan his site for safety flaws. In just 15 minutes, the AI found 13 security issues, such as outdated software and unsafe connection settings. The AI then took about 1 hour to completely fix all of those problems.
In the end, keeping our digital world safe is becoming a fast-paced race. While AI-powered hackers are getting smarter, the tools to defend our data are growing just as quickly. Staying ahead will simply require companies to adopt these smart defense tools before the bad guys do.
