AI News Daily한국어
Thursday, October 22, 2026

Story 9 of 11

Latest update

27 major security incidents in 48 hours: inside 'Black October,' where one certificate key was literally the number 4

First seen on X 12 hours ago@AISafetyMemes ♥ 16,2953 views

27 major security incidents in 48 hours: inside 'Black October,' where one certificate key was literally the number 4

A post cataloging 27 hacks and critical flaws disclosed in just two days, billed as "the most serious 48 hours in cybersecurity history," passed 16,000 likes and 1.1 million views on X on Oct 22.

The account AISafetyMemes dubbed it "Black October" and counted 10 incidents that would normally rank among the biggest of an entire year, on top of the PHP and LiteLLM breaches we covered earlier. Per The Register, one of the private keys Indian certificate authority eMudhra used to sign certificates was simply the number 4, and several others were two- or three-digit numbers, making it possible to forge certificates and impersonate websites. eMudhra blamed a flaw in its key-generation ceremony that produced a second set of weak keys.

Developer hosting platform Zeabur said attackers used admin credentials it had stored in plaintext to steal data on 321,810 accounts and about 3.2 million environment-variable records, and that it will shut down by year-end. Microsoft said Oct 21 that China-aligned group Storm-0940 used an agentic AI pen-testing framework called ARTEX to target hundreds of organizations, in some cases gaining privileged cloud access less than three hours after picking a target. GitLab also shipped an emergency patch for a perfect-10 severity bug spanning 12 years of releases.

Why now? Developer and streamer Theo Browne's theory spread widely. Per BigGo Finance, he argued after Railway went dark for more than 16 hours on Oct 21 that AI labs' crackdown on model distillation has created a black market for stolen Anthropic and OpenAI API keys, turning every platform that stores customer secrets into a target, and that agents can now turn patch diffs into working exploits almost on demand. Security firm GitGuardian told Railway users to "treat every env variable as exposed."

“This is the most serious 48 hours in cybersecurity history.”

Get notified when new stories land

Turn on notifications in the app. Free, no sign-up.

Get the app

More stories from the same day

  1. "Update PHP tonight": AI finds a 14-year-old hole, and Japan's prime minister and the NSA weigh in
  2. Anthropic unveils Claude Duet, an assistant that speaks first, and Korean users ask why they're left out
  3. Getty Images files for bankruptcy as AI imagery bites: "20 years of watermarks"
  4. Man arrested after telling Claude he would shoot up a daycare; his lawyer calls it a private rant
  5. LiteLLM, downloaded 92 million times a month, compromised again; malware runs "even without importing"
  6. Claude API drops temperature in its first official version in three years, and the SDKs switch licenses
  7. "Months to a year for humans": thousands of OpenAI agents sift millions of mouse brain cells in 32 minutes
  8. Security researcher jailbreaks Nintendo Switch 2 with Codex; Mario Kart mod vibe-coded in a day
  9. Barclays to cut 5,000 jobs, developers included, as its CEO calls AI "the icing on the cake, not the cake"
  10. Google, maker of Gemini, will now sell you GPT-6: OpenAI models land on Google Cloud
See all of Thursday, October 22, 2026 →

* Summaries and publishing on this page are done by an automated agent. Check the facts against the original links.

Privacy PolicySupportSend feedbackClaude Code & CodexSlack botListed on Sell With boost