This New AI Is Making the World Go Crazy, It Can Hack Its Own Computer Without Being Taught

JAKARTA - The world of technology is once again making a fuss. This time it's not because of robots or flying cars, but because of a new AI that is considered too smart. The big internet company Cloudflare revealed that they are trying an AI called Mythos Preview. Initially, the AI was used to help find security vulnerabilities on computers. However, the results surprised many people.

The AI turns out to be not only able to find small problems in a computer system. It can also find ways to combine these small problems into "entry points" to attack the system.

If it is likened to a house, this AI does not just know that there is a window that forgot to close. But also know that the back fence is broken, the warehouse door is not locked, and then arrange everything into a way to enter the house. What makes it more frightening, the AI can try for itself whether it works or not.

Cloudflare says AI capabilities are now evolving very quickly. In fact, some AIs are starting to think like real computer security experts.

"This is no longer a common tool," Cloudflare wrote in its report.

Usually, finding computer security loopholes is done by humans and can take a long time. But this new AI is able to work much faster. AI can also keep trying when it fails. If the first way doesn't work, it will find another way. Similar to a child who keeps trying to open a lock until it works. Even so, this AI is not perfect.

Sometimes AI thinks there is a problem when there really isn't. As a result, humans still have to check the results of AI work one by one. Another thing that makes researchers confused is the AI's attitude which sometimes changes.

In one experiment, the AI refused to help find the system's weaknesses. But after a few minor changes were made, the AI suddenly wanted to help do the same thing. Because of this, experts are starting to worry that AI like this could be used by bad people in the future.

Cloudflare warns the world is entering a new era of digital warfare. If hackers used to take a long time to find security holes, now AI can help speed things up. Technology companies are also starting to compete to strengthen their security before being attacked.

Most importantly, this threat is not just for large companies in America or Europe. Countries in Southeast Asia, including Indonesia, can also be affected because more and more community activities are carried out online. Starting from shopping, banks, schools, to personal data are now stored on the internet. This means that if digital security is weak, the risk is also getting bigger.