Are we looking at the same charts? To me the rise started a bit before before, which I'd index on either agentic coding getting wide spread adoption (seeing their PR frequency chart lines up with that) or the ai models are really good at hacking btw blog posts raising awareness and having people try more.
There are plots later in the article where a renewed spike lines up much better with 5.3 release, indicating general ability to find issues increasing (if I understood correctly that the right hand chart is announced 0days)
But to me this tracks a lot more with "we should slow down the slop machine and have humans who care patch the bugs the improved fuzzers and scanners are finding" than other interpretations
The charts are pretty scary, seems like the number of cyberattacks went up dramatically after GLM5.3 was released.
Are we looking at the same charts? To me the rise started a bit before before, which I'd index on either agentic coding getting wide spread adoption (seeing their PR frequency chart lines up with that) or the ai models are really good at hacking btw blog posts raising awareness and having people try more.
There are plots later in the article where a renewed spike lines up much better with 5.3 release, indicating general ability to find issues increasing (if I understood correctly that the right hand chart is announced 0days)
But to me this tracks a lot more with "we should slow down the slop machine and have humans who care patch the bugs the improved fuzzers and scanners are finding" than other interpretations
This link resolves to just https://www.bridgewater.com/ after dismissing the disclaimer.
Is the disclaimer the point, or...?
edit: closing the disclaimer makes it act like an adult website to redirects to the 'safe' homepage. Clicking agree loads the article.
Yeah. Clicking on the link the second time opens the actual article. Not sure why. Sorry.