Will AI Hackers Replace Human Ones?

AI hackers may not stand on two feet yet, smoke cigarettes, or sit in front of computers, but be forewarned: AI hackers are taking shape and are helping direct attacks on American corporations and government institutions.

Artificial Intelligence is the latest and greatest breakthrough, having arrived on the scene in recent years and taken hold. The public, small businesses, and organizations have embraced the technology and possibilities. Everyone seems to love AI.

That’s the good news. The bad news is that hackers, especially foreign-based cyberthieves, have taken the technology and developed ways to turn it against U.S. companies and businesses. In this case, it’s suspected that hackers in China, sponsored by their government, were behind the attack.

This is beyond hackers using technology to write perfect emails or create fake photos and videos. This goes beyond that. The hackers used AI to automate their attacks fully.

Anthropic AI in the hands of hackers.

The hackers used an AI platform from Anthropic AI. They’re the company that developed Claude (a competitor of ChatGPT), defined as a “next-generation AI assistant based on Anthropic’s research into training helpful, honest, and harmless AI systems.”

Well, so much for good intentions.

According to a story in the Wall Street Journal, the Chinese-backed hackers took full advantage of the power and potential of Anthropic’s AI to do the following:

• Automated a hacking campaign that broke into the networks of foreign governments and large corporations.
• Design the attacks so that up to 90% of the effort was automated.
• Stringing together individual hacking efforts into a larger, autonomous, and nearly seamless hacking campaign.

According to a representative of Anthropic, there was very little human interaction during the hacking venture, which broke into around four organizations that we know about. The representative was quoted in the WSJ as saying the attacks were managed “literally with the click of a button, and then with minimal human interaction.”

Fortunately, the hacking didn’t go unnoticed. Anthropic itself was able to spot what was happening, block the hackers’ account, and end the attacks before more networks were compromised. They also said that up to 30 corporations and governments had been targeted. Anthropic added that, based on the clues they saw, it had all the earmarks of a Chinese-designed effort.

Will AI hackers be the new trend?

Is this the new direction of hacking, nearly fully automated campaigns, powered by

AI? It might be.  Google reported that Russian hacking attacks against Ukraine used Artificial Intelligence to create customized malware on the spot, in real time.

According to the WSJ, in the summer of 2025, a cybersecurity firm noticed Chinese-backed hackers using AI to automate attacks against a variety of organizations. The cybercriminals engaged AI to help them do the following:

• Find out what organizations to target.
• Write and deliver effective emails to deceive key employees.
• Create and write malware that would infect networks upon launch.

With AI tools, such as Claude, hackers can now handle complex tasks faster and more effectively.

Countries are pointing fingers at each other.

The U.S. Government has been saying for several years that China is targeting U.S.-developed AI technology for exploitation. They’ve claimed China will use AI to infiltrate U.S. companies and government departments to steal data.

In response, China said that, after recent suspected AI-powered hacking campaigns, that’s not the case. It accused the U.S. of running a slander campaign to make China look like the bad guy. A spokesman for the country, based in Washington, D.C., says it is cracking down on cyberattacks.

How can hackers use AI so effectively?

The problem with AI might simply be its availability to everyone, its ease of use, and its incredible power and potential.

Any Artificial Intelligence company knows the power of its products cuts both ways, for good and not-so-good. When it comes to cybersecurity, Anthropic reportedly said it hopes to develop AI tools that “supercharge” cybersecurity defense systems. Of course, they also know that powerful AI tools in the hands of attackers (and what they might do) is a reality too.

In fact, an Anthropic manager in the defense area, Logan Graham, stated the conflict clearly. He runs a team that runs tests anticipating “catastrophic risks.” 

The company is hard at work creating new skills for its AI that help defenders succeed, but not hackers.

There’s even a term for that strategy—it’s called “vulnerability discovery.”

As quoted in the Wall Street Journal, Logan said, “These kinds of (defense) tools will just speed up things up. If we don’t enable defenders to have a very substantial permanent advantage, I’m concerned that we maybe lose this race.”

Follow the Easy Prey Podcast for more about AI.

Chris Parker, CEO of WhatIsMyIPAddress.com, hosts the Easy Prey Podcast, which covers a full range of critical topics, including AI and cybersecurity. If you click the link, you can search for the topic that interests you or look at the extensive list of past episodes.

You can also find more resources in our Learning Center at WhatIsMyIPAddress.com, including articles, tools, and links to podcast transcripts and interviews.