Hackers are using artificial intelligence to automate a massive cyber attack, and that’s very bad news for all of us.

Now I see it: Terminator returns in 2021, then casually walks past the offices of Boston Dynamics, Tesla, even X1 and Figure AI, and stops in front of Anthropic instead. In his signature Austrian accent, the Terminator flexes his enormous muscles and says, “I have to stop the programmers from doing this.” » DO Claude AI I City Prevent the first large-scale, almost fully automated cyberattack in 2025.” Meanwhile, robot developers are on the run, believing that the Terminator may go into hiding again within a decade.

There is no real Terminator, but there are extremely worrying signals about the rapid development of artificial intelligence and new concerns about his weapons. This week, Anthropic announced that it had largely foiled a “massive AI-orchestrated cyberespionage campaign.”

The alleged attack, which may have been carried out by Chinese hackers in September this year, targeted major technology companies, financial institutions, chemical companies and government agencies. Any of these attack vectors should give you pause, especially those that target the average consumer. Government agencies can mean almost anything, including infrastructure, water control systems, electricity and even food safety.

It’s an attack that, according to Claude AI’s Anthropic, couldn’t have happened just a year ago. This really doesn’t surprise me. As I like to say, we now live in the age of artificial intelligence, where the pace of development and innovation is three times faster than in previous eras of technological innovation. If Moore’s Law required a doubling of the number of transistors in a processor every 18 months, a doubling of intelligence in LLM development could occur every six months.

IF Anthropic explains in a blog postArtificial intelligence models:

• They are smarter now
• Have the ability to take autonomous actions, link them together and even make decisions with little human intervention.
• You can also use tools to search the Internet and recover data on your behalf.

Hackers using artificial intelligence to further their efforts is nothing new. The spam, text messages and phone calls you receive every day are also increasing as AI makes it easier to develop new identifiers and strategies.

But this latest development appears to be helping hackers launch large-scale attacks with little more than basic programming and, more importantly, instructions.

According to Anthropic, “Overall, the threat actor was able to execute between 80% and 90% of the campaign using AI, with only sporadic human intervention.”

The only good news is that Anthropic acquired the company and quickly shut it down before it had made enough progress to have much impact in the real world.

The next scary level

However, these pirates were highly motivated and quite intelligent. They bypassed Anthropic’s defenses by breaking the attack into small, harmless pieces, tasks that individually seemed harmless, but together made up the whole attack.

In my opinion, everything Anthropic has shared about this cyber attack is very concerning and should be taken as a warning to all of us.

The rapid development of artificial intelligence is making these platforms increasingly smarter. In particular, AI agents (models that can perform tasks on your behalf) are at the forefront of the development of virtually all AI platforms, including Google’s (see SIMA2that can play autonomously in virtual worlds) and OpenAI, and while most are used for good purposes, these features are clearly tempting for cyber attackers.

It may seem like this is only a concern for governments, businesses and infrastructure, but failures in any of these systems or businesses can often result in loss of services, support, resources and protection for consumers.

So yes, Skynet is still the fictional big bad of our AI nightmares, but it doesn’t take an army of robots to take down society, just one or two hackers with access to the best of AI. The next Terminator will surely be the first to visit companies with artificial intelligence.