Weeknotes 2026.40
Posted on Do 01 Oktober 2026 in Blog
It was kind of a short week. On Monday and Tuesday we had a team building event with a little hike and a campfire in the evening. It is always good when we all come together outside of the office.
The weather was good, and we had a lot of fun. Overall, it was a great event.
On the other hand, a two-day team event means that there are only three days left and that is usually a little bit stressful.
Data leak at Microsoft
faav has written how he found a vulnerability in Microsoft's Titan system and could've accessed 17 trillion records. It is an interesting writeup.
How I Could've Accessed 17 Trillion Microsoft Records
Kernel Recipes 2026
Greg Kroah-Hartman has presented a talk about security and the findings of LLMs at the Kernel Recipes 2026 conference. And Andres Freund presented some Postgres challenges at the Kernel Recipes 2026 conference and how the Postgres community interacts with the Linux kernel.
QGIS UI
Anita Graser has written about ribbon UI experiments in QGIS.
QGIS tabbed ribbon UI experiments
LLLM: Gemini 4 and Kolibri released
Google has released Gemini 4. According to some benchmarks, Google caught up with the top LLM models.
- Gemini 4 Argon: Google is back as one of the top three labs in intelligence achieved
- Google unveils Gemini 4 Argon, retaking benchmark lead over OpenAI and Anthropic — but in limited release
Aleph Alpha released Kolibri, a LLM with a focus on the German language.
Aleph Alpha Kolibri: How the Sovereign German LLM Works
Data breach ticker
- Another week, another data breach for Revolut customers
- Russian pizza chain with 1,500 locations confirms cyberattack following hacker claims
- Pentagon data breach of military personnel raises national security concerns
- Over 16,000 Supabase databases expose PII, passwords, auth tokens
- Over 543,000 valid credentials exposed in public GitHub repositories
- Times Car confirms data breach affecting 6.6 million user accounts
- Cyberattack on major Polish invoicing platform exposes customer data