<h2>Must know</h2>
<ul>
<li><strong>OpenAI goes all-in on ads, privacy concerns mount</strong>: OpenAI has launched ChatGPT Go, an $8/month subscription tier with ads, rolling out globally after an India debut. This aggressive monetization strategy, detailed by sources like ‘The A in AGI stands for Ads,’ is a direct response to Google Gemini’s market share gains. While OpenAI projects massive ad revenue, critics are raising serious privacy and ethical concerns about using user conversation data for ad targeting, especially given the company’s assurances that user conversations remain private from advertisers. <a href="https://app.daily.dev/feed-by-ids?id=ArnlILqap&amp;id=ep7p1hnQy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI code generation still struggles with basic memory bugs</strong>: Despite rapid advancements, AI code generation tools consistently reproduce a specific memory management bug involving buffer allocation, as highlighted by ‘Algorithmic Influence and Persistent AI Code Generation Flaws.’ This flaw, where AI returns allocated buffer size instead of actual data, can lead to security vulnerabilities by exposing uninitialized memory. The persistence of this bug suggests a fundamental limitation: LLMs tend to statistically mirror patterns, even erroneous ones, from their training data, indicating that current statistical approaches may not easily eliminate such mistakes. <a href="https://app.daily.dev/feed-by-ids?id=mY0RAoEVz&amp;id=x9XQt7sEo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>China’s massive satellite internet ambitions challenge Starlink</strong>: China has filed plans with the ITU to launch two mega-constellations, CTC-1 and CTC-2, totaling nearly 200,000 satellites—four times the size of Starlink. This move, reported by multiple sources, signals China’s intent to dominate orbital radio spectrum and could significantly alter global internet access and security. Meanwhile, Iran successfully disrupted Starlink internet using military-grade GPS jamming, raising concerns about the system’s resilience and its reputation as censorship-resistant in conflict zones. <a href="https://app.daily.dev/feed-by-ids?id=9Dqo8zul1&amp;id=VVK9EFJ8w" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Figma now controllable by AI agents via CLI</strong>: A new command-line tool, figma-use, provides full read/write access to Figma, allowing AI agents to programmatically create shapes, text, components, and styles using JSX syntax, optimized for token efficiency. <a href="https://app.daily.dev/posts/DfAoRLrdU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google’s Antigravity expands to compete with Anthropic’s Co-work</strong>: Google’s Antigravity is evolving beyond coding to include productivity features like creating presentations and organizing files directly from the IDE, leveraging Gemini 3 Flash to compete with Anthropic’s Co-work. <a href="https://app.daily.dev/posts/HVlI4ECqg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered low-code platform for LLM workflows</strong>: fluentAI is an open-source, low-code/no-code platform for building custom LLM workflows visually in the browser, inspired by Node-RED, using ReactFlow and Supabase for serverless execution. <a href="https://app.daily.dev/posts/T2Zh09ohU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pure C image generation model runs without Python or CUDA</strong>: A pure C implementation of the FLUX.2-klein-4B image generation model, antirez/flux2.c, runs without Python, PyTorch, or CUDA, supporting text-to-image and image-to-image generation with optional BLAS and Metal GPU acceleration. <a href="https://app.daily.dev/posts/NsHswJDic" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wine 11.0 stable release brings major architectural improvements</strong>: Wine 11.0 stable is out with 6,300 code changes, including a new NTSYNC kernel module, a completed WoW64 architecture overhaul, and EGL as the default backend for OpenGL, enhancing Windows app compatibility on Linux. <a href="https://app.daily.dev/posts/5LDn1X8et" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wine patches enable direct Adobe Photoshop installation on Linux</strong>: New Wine patches allow the Adobe Creative Cloud installer to function on Linux, enabling direct installation of Adobe Photoshop 2021 and 2025 by fixing XML parsing issues related to legacy Internet Explorer dependencies. <a href="https://app.daily.dev/feed-by-ids?id=WW9lhhDTb&amp;id=jcudSc8Dd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Jakarta EE 12 development and community events update</strong>: The Jakarta EE Platform project is progressing towards Milestone 2 for Jakarta EE 12, targeting January 27, 2026, with JakartaOne Livestream 2025 videos now public and multiple community conferences scheduled. <a href="https://app.daily.dev/posts/YrhkKGKoN" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>The internet is dead, long live the bots</strong>: Apparently, the ‘Dead Internet Theory’ suggests that since 2016, most online content and interactions are AI-generated, not human. This ‘insight,’ sparked by a HackerNews discussion, points to AI-generated code and comments on open-source projects as proof. Because, you know, it’s totally impossible for humans to just be boring and repetitive; it <em>must</em> be the bots. <a href="https://app.daily.dev/feed-by-ids?id=JjrjV7tSv&amp;id=ZeB3UcRxG&amp;id=alpyX7cTa&amp;id=m1Q81Q3xQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code: faster, but still needs a babysitter</strong>: AI-powered code generation is accelerating software development, delivering outputs in minutes instead of days. But don’t get too excited; human review is still ‘essential’ for correctness, security, and architectural compliance. So, the bottleneck just moved from writing code to reviewing the AI’s questionable output. Progress! <a href="https://app.daily.dev/feed-by-ids?id=JjrjV7tSv&amp;id=ZeB3UcRxG&amp;id=alpyX7cTa&amp;id=m1Q81Q3xQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are making us question the meaning of ‘joy’</strong>: A developer, after 20 years of Vim/Neovim, is now ‘good enough’ with AI coding agents to question if learning fundamental tools like awk and sed still makes sense. The piece explores the ‘tension between efficiency gains and the loss of joy’ that came from the craft of coding. Because nothing says ‘technological advancement’ like replacing human satisfaction with existential dread. <a href="https://app.daily.dev/posts/vFzyyNK3z" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
