
What happened
France, Denmark, Finland, and the Netherlands took government websites offline Wednesday after developers used AI to identify a bug affecting up to 70% of the world's websites, which Google's Wiz co-founder called "the most severe AI vulnerability… ever found".
Why it matters
Fixes are now public, and a researcher told Reuters hackers can use AI to reverse engineer the bug in a "race to exploit" it, with the next step a "mass compromise of every vulnerable web server on the internet".
WHO IT HITSGovernment website operators and IT teams running public-facing web servers are the most exposed, since fixes are already public and the researcher warns of a race to exploit them.
Summaries like this, in your inbox every morning.
The discovery flips the usual narrative around AI and security. Rather than automating attacks, AI here surfaced a flaw so widespread that four governments — France, Denmark, Finland, and the Netherlands — chose to pull their websites offline on Wednesday rather than wait. The co-founder of Google's Wiz team told The Wall Street Journal that humans probably could not have found it without AI, which suggests the tool is becoming a genuine detection aid rather than only a threat.
The same capability cuts both ways. Once fixes are public, a cybersecurity researcher told Reuters, AI can help attackers reverse engineer the bug in a "race to exploit" it, with the stated next step being "mass compromise of every vulnerable web server on the internet." That puts the emphasis on speed of patching: Japan's prime minister urged "everyone to take action tonight," a signal that the window between disclosure and exploitation may be measured in hours rather than weeks. The episode is likely to sharpen debate over how quickly AI-found vulnerabilities should be disclosed, and to whom.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
PHP developers released urgent fixes on Wednesday for critical vulnerabilities that had lurked in the widely u…

Elon Litman showed that a loss term built from (2^30 + u) − 2^30 reads 0.0 in float32 while PyTorch's autograd…

The NSA warned that two critical PHP vulnerabilities, present since version 5 in 2004, affect roughly 70% of w…

Hugging Face CEO publicly criticized Netflix's AI hack documentary, saying it gets the story wrong

CrowdStrike Counter Adversary Operations uncovered four previously unreported threat actors weaponizing AI age…

Palo Alto Networks said AI Runtime Security on NVIDIA BlueField DPUs is now generally available across AI fact…
