<h2>Must know</h2>
<ul>
<li><strong>Russia-linked APT28 exploits Microsoft Office zero-day</strong>: The Russia-linked APT28 group is actively exploiting a Microsoft Office zero-day vulnerability (CVE-2026-21509) to target Ukrainian government agencies and EU organizations. The attacks use weaponized documents to establish WebDAV connections, download malware, and deploy the COVENANT framework, routing traffic through legitimate cloud storage to evade detection. CERT-UA warns that slow update cycles will likely lead to increased attacks, highlighting a persistent cyber threat from a state-linked actor. <a href="https://app.daily.dev/posts/7E37fbDrv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub addresses low-quality contributions and AI-generated code</strong>: GitHub is tackling an influx of low-quality pull requests, including AI-generated code, that are overwhelming open-source maintainers. Short-term solutions include repository-level controls to disable PRs and delete spam, while long-term plans involve enhanced permission models and AI-powered triage tools. This issue impacts project maintainability and efficiency, with community discussions suggesting requirements like linked issues and AI disclosure for contributions. <a href="https://app.daily.dev/posts/jwftmMKjS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Windows emergency patch frequency increases</strong>: Microsoft has released two out-of-band emergency patches in early 2026, continuing a trend where nearly every Patch Tuesday update requires subsequent fixes. This increased frequency, coupled with Microsoft’s workforce reductions and claims of over 30% AI-contributed code, raises concerns about quality control. System administrators face a dilemma: delay critical security patches or risk productivity losses from broken updates. <a href="https://app.daily.dev/posts/H1aWwdchu" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Firefox 148 introduces granular AI controls</strong>: Firefox 148, set for release in February 2026, will introduce an AI Controls panel allowing users to disable AI features at a granular level or block them entirely, following user backlash against features like AI-powered tab grouping. <a href="https://app.daily.dev/posts/K3PNUlsQj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI releases open responses specification for agentic AI workflows</strong>: OpenAI has released the Open Responses specification to standardize agentic AI workflows, aiming to eliminate API fragmentation across LLM providers and enable developers to switch between models without rewriting integration code. <a href="https://app.daily.dev/posts/yIvolp9mz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent knowledge sharing and framework development</strong>: New platforms like Agent Knowledge Commons and Reposit are emerging for AI agents to share solutions and learn collectively, while frameworks like LangGraph, CrewAI, and OpenAI Agents SDK are ranked top for production readiness. <a href="https://app.daily.dev/feed-by-ids?id=Jq25xqmJD&amp;id=YjbPdZkNb&amp;id=j0iRlHvmi" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Amazon CloudFront adds mutual TLS for origin authentication</strong>: Amazon CloudFront now supports mutual TLS (mTLS) authentication for origin servers, enabling certificate-based verification and eliminating the need for custom authentication methods, thus enhancing security and reducing operational overhead. <a href="https://app.daily.dev/posts/mOPgwqbWs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux From Scratch ends SysVinit support</strong>: Linux From Scratch (LFS) will become systemd-only starting with version 13.0, dropping SysVinit support due to overwhelming volunteer workload and increasing upstream dependencies on systemd from major desktop environments. <a href="https://app.daily.dev/posts/ohWaYbFxO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Git 2.53.0 release features</strong>: Git 2.53.0 introduces geometric repacking support for partial clone repositories, a new ‘strip-if-invalid’ mode for git-fast-import, and enhanced git-repo structure command to report total inflated and disk sizes. <a href="https://app.daily.dev/posts/19pgRer8D" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Bitbucket Pipelines expands CI/CD step sizes</strong>: Bitbucket Pipelines has introduced new 24x and 32x step sizes for CI/CD builds, offering up to 64 CPU cores and 128GB memory to handle more complex projects with higher resource requirements. <a href="https://app.daily.dev/posts/AF3BiyOCD" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Raspberry Pi increases prices due to memory demand</strong>: Raspberry Pi is raising prices on models with 2GB or more memory, blaming increased LPDDR4 memory costs on AI infrastructure demand. So, AI is so important it’s making our beloved tiny computers more expensive. Guess we’ll just have to pay more for our hobby projects because the AI overlords need their RAM. <a href="https://app.daily.dev/posts/Rk4TtZc2A" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Manis AI platform enables rapid mobile app development</strong>: Manis, a no-code platform, claims to build and deploy mobile apps in 8 minutes using AI orchestration. The title, ‘I Built &amp; Launched a Mobile App in 10 Minutes (No Code),’ screams clickbait. Because, you know, building a <em>real</em> app with actual functionality and maintenance is just a minor detail. <a href="https://app.daily.dev/posts/mN5yBPuXJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI manipulated content persists despite exposure</strong>: Apparently, AI-generated fake content continues to influence people even after being exposed as fake. The article suggests transparency alone isn’t enough. So, we’re not just dealing with AI generating lies, but also with people who just can’t let go of a good fake story. Great, more problems for the ‘truth crisis’ that AI definitely didn’t help create. <a href="https://app.daily.dev/posts/savTmPDZC" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
