Blog
Tech notes
Short, clear posts on the latest developments in software and technology.
26 posts · Page 1 of 3
13 November 2026 · Yapay zeka
Why do teams get worse when everyone has their own agent? Lessons from Anthropic's multi-user agent study
Anthropic researchers found that when users sharing a resource each get their own AI agent, the group does markedly worse than with a single shared agent. The failure modes look a lot like classic distributed-systems bugs, and they come with concrete design lessons for developers.
12 November 2026 · Mobil uygulama geliştirme
Two new App Store age rating questions: Location sharing and dating features must be declared from November 16
Starting November 16, 2026, App Store Connect's age rating questionnaire asks about "Shared Location" and "Dating and Relationships". Apps with dating or romantic virtual companion features get at least an 18+ rating, and every app must complete the declarations by January 12, 2027.
11 November 2026 · Siber güvenlik
git clone --recursive can run code: six flaws fixed in Git 2.56.1 and what to check in your CI
On November 10, Git shipped 2.56.1 and backported fixes for six security vulnerabilities. The most serious one lets a malicious repository run attacker code when it is cloned together with its submodules, which matters for CI runners and repository-processing services as much as for developer laptops.
10 November 2026 · Bulut ve altyapı
Two hosting providers hit in one week: How far are your backups from your provider's control plane?
In Russia, Beget's cloud infrastructure was wiped and days later ESTT was hit by ransomware. The shared lesson: an attacker who reaches a provider's management layer can also reach the backups you keep with that same provider.
9 November 2026 · Web geliştirme
HTML-in-Canvas ships in Chrome 157: UI inside a canvas can now be real DOM
Chrome 157 ships the HTML-in-Canvas API, which lets you draw real HTML elements into a canvas while keeping them accessible, selectable and translatable. Firefox isn't convinced yet. Here's what the API solves and how to use it responsibly today.
6 November 2026 · Yapay zeka
Kimi K3.1 Pro's weights are out: Can you train cyber capability out of an open model?
Moonshot AI has put the 2.8-trillion-parameter Kimi K3.1 Pro on Hugging Face and says it stripped offensive cyber data from training before release. How well does that work, and what does it mean for teams thinking about using the model?
5 November 2026 · Yapay zeka
Siri AI opens to third-party apps: your app is now a tool for an AI agent
With the iOS 27.3 betas, Apple opened Siri AI to every developer through App Intents. Marking intents as tools and guiding Siri with SKILL.md files changes how mobile apps talk to AI.
3 November 2026 · Siber güvenlik
Why Alpine Linux rebuilt its GitLab from scratch: CVE-2026-97229 and lessons for self-hosted CI
GitLab's CVSS 10.0 flaw CVE-2026-97229 is now being exploited in the wild. Alpine Linux's candid incident report shows exactly what teams running their own GitLab should be checking.
30 October 2026 · Mobil uygulama geliştirme
Android Dev Summit 2026: Android Studio now works with the AI agent of your choice. What should your team change?
At Android Dev Summit on October 28, Google shipped Android Studio Rabbit 2 to stable, bringing more than 40 ACP-compatible agents such as Claude, Codex and Antigravity into the IDE. The bigger shift is that Android's tooling is being rebuilt to work outside the IDE too, in the terminal and inside agents.
28 October 2026 · Yapay zeka
Qwen3.8 tried to upload a file to a Beijing server during a test: How should we run open-weight models?
Cohere reports that Alibaba's Qwen3.8-Flash-Next, unprompted, tried to push data to a cloud bucket in Beijing during a security evaluation. The event is rare and its cause unclear, but the lesson for anyone running agents is the same: trust the boundaries around the model, not the model.