Cybersecurity in the news, but actually AI is bringing us the future of less security holes.

I've been getting the list of fixed Linux kernel CVEs every week for now six weeks. https://preview.redd.it/js35gdpys8xh1.png?width=553&format=png&auto=webp&s=fb3e56fa5953a724004d1cc44cb15d867150468d In this chart is axis 'Intro date' meaning the week the security hole was actually originally made into released code. Fetch date is the week I fetched the date's of public CVEs. What I've seen from history data, 50% of CVE's are made public after 2 months and 100% after three months. The gray areas are that. Normally when advance a week, the amount of hole's should follow, week after week. The gray area simple says the data in that is not yet secure. But it seems big drop of new CVE's was in 20.4. Then in 15.6 and after 10.8 there hasn't been but 2 new CVE's. In the same time there have been fixes of _old_ holes, but it seems introduction of new holes is almost stopped. The 21.10 and 22.10 fetches are because there was a fault in my script that it didn't get all dates correctly and running it fixed in 22.10 got additional 600 CVE's, but as you can see from numbers, that was mostly old security holes. Btw, those last 2 CVEs were found in 26.9. and fixed in 28.9. What I see from that da