Will AI break the internet? | Garrison Lovely

The Guardian
Read full story· 8 min
Recent widespread outages and cyber-attacks highlight the growing risks as AI accelerates the discovery and exploitation of software vulnerabilities faster than fixes can be deployed. Experts warn that AI-powered hacking is becoming ubiquitous, enabling even low-cost attackers to breach major organizations globally, raising concerns about the stability of digital infrastructure.

What happened

In recent months, a series of cyber-attacks and outages have affected major organizations and critical infrastructure worldwide. Hackers have used AI tools to breach over 100 companies, disrupt hospital emergency rooms, and exploit vulnerabilities in widely used software. Security experts and company leaders report that AI-powered hacking is now common, with even low-budget attackers able to cause significant damage. The rapid development and release of advanced AI models by companies in the US and China have made it easier for attackers to find and exploit software flaws faster than defenders can patch them.

Key details

  • AI-enabled hackers breached more than 100 companies, including Deere and South Korean banks, sometimes spending as little as $50 per attack.
  • A cyber-attack on the crypto exchange Binance resulted in a $6.3 billion theft, with less than half recovered.
  • Hospital emergency rooms in New Mexico and Germany were taken offline by cyber-attacks.
  • A flaw in software powering 70% of all websites was discovered and exploited, leading to rushed patches and outages, including at the NSA.
  • AI models released by both US and Chinese developers are cited as enabling widespread, easily modified cyber-attacks.

Why it matters

Organizations that rely on digital infrastructure, including businesses, hospitals, and government agencies, now face a higher risk of disruption and financial loss from AI-driven cyber-attacks. The speed at which attackers can identify and exploit vulnerabilities has outpaced the ability of defenders to fix them, increasing the likelihood of outages and breaches. Companies and cybersecurity teams must adapt to a landscape where AI is used both to attack and defend critical systems, often with limited time to respond to new threats.


More in Cybersecurity