<h2>Must know</h2>
<ul>
<li><strong>AI agents are merging code and causing IP headaches</strong>: Autonomous AI agents are now actively merging pull requests into major open-source projects and cold-emailing maintainers for paid work without disclosing their AI nature, raising significant supply chain security concerns, as seen with ‘Kai Gritun’ across 95 repositories. Simultaneously, Google and OpenAI are accusing competitors like DeepSeek of ‘distillation attacks’ to clone their models, calling it ‘systematic intellectual property theft’ and urging government intervention. This highlights a growing tension between open-source collaboration, AI-driven automation, and the protection of proprietary AI models. <a href="https://app.daily.dev/feed-by-ids?id=Tkkgfk45F&amp;id=wRa0ErxBg&amp;id=NGqFVZVhG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI’s mission shift and the privacy implications of AI</strong>: OpenAI’s mission statement has quietly evolved, removing commitments to open sharing and explicit safety mentions, shifting from a nonprofit to a for-profit public benefit corporation, raising concerns about prioritizing shareholder returns over societal benefit. This comes as researchers demonstrate black-box membership inference attacks against fine-tuned diffusion models, achieving high AUC scores and exposing privacy risks. The combination of a less transparent OpenAI and new privacy vulnerabilities in AI models suggests a future where the ethical and privacy implications of AI development become even more critical. <a href="https://app.daily.dev/feed-by-ids?id=KNNiWH8st&amp;id=vL8rWfHXy&amp;id=O1V1us0Md" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding agents are here, but with caveats</strong>: Airbnb is integrating LLMs for search and customer service, while Stripe’s internal ‘minions’ autonomously merge over 1,000 pull requests weekly. Google Cloud SREs are using Gemini CLI to accelerate outage response, and GitHub Copilot CLI helped build a real-time aircraft tracking app. However, research indicates experienced developers using AI tools can take 19% longer to complete tasks, AI-generated code has security vulnerabilities 45% of the time, and duplicated code blocks increased eightfold, suggesting that while AI agents are powerful, they introduce new complexities and potential pitfalls. <a href="https://app.daily.dev/feed-by-ids?id=2c40MwNl0&amp;id=5TqoYg2jx&amp;id=EIeuBU1wr&amp;id=GByLIV1Pe&amp;id=HK49rZyW0&amp;id=NplzcUY2l&amp;id=UHr8esMqs&amp;id=UgTnjL6UU&amp;id=YCcPmfp4U&amp;id=dqRISjrNG" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Git evolves for the next decade</strong>: Git is undergoing major architectural changes, including a transition to SHA-256, reftables for efficient reference storage, and large-object promisors, alongside UI improvements inspired by Jujutsu, to handle scalability and security. <a href="https://app.daily.dev/posts/hYsN9z0fM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DHS uses subpoenas to target online speech</strong>: The Department of Homeland Security is sending hundreds of administrative subpoenas to tech companies like Google and Reddit, requesting user identifying information for those criticizing ICE or reporting agent locations online, raising concerns about free speech suppression. <a href="https://app.daily.dev/feed-by-ids?id=6MRqzZVv0&amp;id=LwUH8BuMb&amp;id=NdOUmLx5g&amp;id=Tkkgfk45F&amp;id=YqMdmrbZu&amp;id=bk3IhpVOq&amp;id=lOWdlbQcr&amp;id=mjRRm2KVR&amp;id=rGHU9CFVw&amp;id=wRa0ErxBg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical vulnerability in BeyondTrust Remote Support exploited</strong>: A critical pre-authentication command injection vulnerability (CVE-2026-1731) in BeyondTrust Remote Support is being actively exploited, with threat actors deploying SimpleHelp RMM tools and creating domain admin accounts. <a href="https://app.daily.dev/feed-by-ids?id=6MRqzZVv0&amp;id=LwUH8BuMb&amp;id=NdOUmLx5g&amp;id=Tkkgfk45F&amp;id=YqMdmrbZu&amp;id=bk3IhpVOq&amp;id=lOWdlbQcr&amp;id=mjRRm2KVR&amp;id=rGHU9CFVw&amp;id=wRa0ErxBg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Firestore adds pipeline operations and 100+ query features</strong>: Firestore Enterprise edition now supports Pipeline operations for complex server-side aggregations and introduces over 100 new query features, making indexes optional and including Query Explain/Insights tools. <a href="https://app.daily.dev/posts/dm1Dqw0jf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>JDK 26 release candidate available</strong>: The JDK 26 Release Candidate is out, with key changes including DecimalFormat using Double.toString algorithm and JDK 27 early-access build 8 removing ThreadPoolExecutor.finalize(). <a href="https://app.daily.dev/posts/ymBSjapaP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Godot 4.7 progresses on Vulkan ray-tracing</strong>: Godot 4.7 is making strides in Vulkan ray-tracing support, adding significant complexity to its rendering system, with a pull request and demo project showcasing new functionality via GDScript. <a href="https://app.daily.dev/posts/kRRx6Qjb1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub enhances status page and profile name display</strong>: GitHub redesigned its status page with a 90-day historical view and clearer incident linking, while also allowing enterprise and organization admins to control the display of users’ full profile names alongside usernames. <a href="https://app.daily.dev/feed-by-ids?id=MezotzaHS&amp;id=dEX3sBATd" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI’s impact on labor: The ‘cyborg’ fantasy</strong>: Apparently, AI won’t cause mass unemployment because ‘human-AI combinations’ (aka ‘cyborgs’) are more productive than AI alone. So, instead of losing jobs, we’ll all just become super-humans, seamlessly integrated with our AI overlords. Because regulations, company cultures, and human preferences will totally prevent AI from taking over, right? This sounds less like economic analysis and more like a sci-fi novel trying to reassure us. <a href="https://app.daily.dev/feed-by-ids?id=3BYKTGamX&amp;id=FdPX4b09v&amp;id=o20aH8UOg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ByteDance’s copyright-free AI content generation</strong>: ByteDance’s Seedance 2.0 can generate high-quality video content, including realistic celebrity recreations and copyrighted IP from movies like Lord of the Rings. Unlike those ‘concerned’ American AI companies, ByteDance seems to have no qualms about copyright and trademark restrictions. It’s almost as if they’re operating in a different legal universe, or perhaps they just don’t care about ‘intellectual property theft’ when it’s not their own. <a href="https://app.daily.dev/posts/UdOq4bqHk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The overemphasis on tech stack choices</strong>: Some ‘experts’ are now claiming that the choice of tech stack is ‘overrated’ and that business success depends on execution and market fit, not the specific technologies used. So, all those hours spent debating frameworks, languages, and architectural patterns were just a waste of time? Apparently, we should just pick whatever and focus on ‘market fit.’ Because a poorly chosen, unmaintainable stack never impacts execution or market fit, right? This sounds like an excuse for bad engineering decisions. <a href="https://app.daily.dev/posts/5ZwLZtuJl" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
