<h2>Must know</h2>
<ul>
<li><strong>Microsoft patches zero-days and AI vulnerabilities</strong>: Microsoft’s February 2026 Patch Tuesday addressed 59 vulnerabilities, including six actively exploited zero-days. Critically, it also patched remote code execution flaws in GitHub Copilot and multiple IDEs (VS Code, Visual Studio, JetBrains) stemming from AI prompt injection vulnerabilities. These AI-related flaws pose significant risks to developers due to potential access to sensitive credentials and infrastructure keys, as reported by Krebs on Security. <a href="https://app.daily.dev/feed-by-ids?id=zk4Rm0PBv&amp;id=BOPPPXswR&amp;id=uxvhPzT2r" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents introduce new security risks</strong>: The rapid deployment of autonomous AI agents, expected in 40% of enterprise applications by 2026, introduces significant security vulnerabilities. Experts warn that AI agent ‘skills’ (often markdown files providing context) are susceptible to supply chain attacks, hidden commands, and hallucinated commands. GitGuardian, which just raised $50M, is focusing on securing these non-human identities, highlighting the urgent need for robust governance and least privilege principles for AI agents. <a href="https://app.daily.dev/feed-by-ids?id=FBoFtrAqF&amp;id=0rQwEgnbB&amp;id=lhjmWCnek&amp;id=UhlBeqTGk&amp;id=i2YjgAdJA&amp;id=zFYu8Wov8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open source sustainability crisis deepens</strong>: The open-source software ecosystem is facing a sustainability crisis, with companies profiting from the unpaid labor of maintainers who are experiencing burnout and lack of support. High-profile incidents like Heartbleed and XZ Utils expose critical dependencies maintained by overworked individuals. Lotte Pitcher at NDC London 2026 emphasized that consumers must reframe support for OSS as risk reduction, not charity, by sponsoring projects and contributing time. <a href="https://app.daily.dev/posts/aIx715a23" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Java adoption for AI development grows</strong>: A recent Azul report indicates Java adoption for AI development jumped from 50% to 62% year-over-year, with JavaML, Deep Java Library, and OpenCL being popular choices, though 81% of users are moving away from Oracle Java due to pricing. <a href="https://app.daily.dev/posts/LDTrFcHSl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches unified data engineering platform</strong>: Databricks introduced Lakeflow on Azure Databricks, a new unified data engineering solution for ingestion, transformation, orchestration, and governance, featuring AI-powered debugging and serverless compute. <a href="https://app.daily.dev/posts/DckO36Vaq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pandas 3.0.0 introduces breaking changes</strong>: Pandas 3.0.0 is out with breaking changes, including a dedicated string dtype, formal Copy-on-Write semantics, and early support for declarative column transformations, requiring Python 3.11+. <a href="https://app.daily.dev/posts/Bf1dK14FP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Erlang/OTP 29.0-rc1 released</strong>: Erlang/OTP 29.0-rc1 is available, featuring native records, multi-valued comprehensions, and quantum-resistant SSH key exchange by default, while dropping 32-bit Windows builds. <a href="https://app.daily.dev/posts/fTCktIrzF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ESLint 10.0.0 removes old config support</strong>: ESLint 10.0.0 will remove support for <code>.eslintrc</code> configuration files, making flat config the sole option and improving config resolution for monorepos. <a href="https://app.daily.dev/posts/eQ0JUyyzE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PyTorch Foundation expands beyond PyTorch</strong>: The PyTorch Foundation, now led by Mark Collier, has expanded to host critical AI infrastructure like vLLM, DeepSpeed, and Ray, aiming to foster open collaboration across the AI ecosystem. <a href="https://app.daily.dev/feed-by-ids?id=2JQZDZ3HD&amp;id=pbGBRZXIy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust web framework Leptos gains traction</strong>: Leptos, a Rust-based web framework created by an Episcopal priest, is gaining significant community adoption with over 19,000 GitHub stars, offering performance comparable to Solid.js. <a href="https://app.daily.dev/feed-by-ids?id=LEUKhS9TX&amp;id=cCyxGk3Tq&amp;id=gkCUaPXMf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI will make you obsolete, unless you buy our tools</strong>: Apparently, AI models like GPT-5.3 Codex and Claude Opus 4.6 are now so advanced they can autonomously complete complex, multi-hour tasks, eliminating the need for human oversight. Industry leaders are predicting 50% of entry-level white-collar jobs will vanish in 1-5 years, so you better immediately adopt paid AI tools and experiment daily, or you’ll miss this ‘closing window for competitive advantage.’ Because nothing says job security like constantly chasing the next shiny AI object. <a href="https://app.daily.dev/posts/lGyFb7XTB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered testing is the death of traditional testing</strong>: Datadog and GitHub Copilot are pushing new AI-powered testing tools, with one article boldly proclaiming ‘The Death of Traditional Testing.’ These tools promise to automatically generate tests, infer code change intent, and ‘auto-heal’ tests when UIs change. So, instead of writing actual tests, we can just let the AI do it, and it’ll magically adapt to all our code changes. What could possibly go wrong with fully automated, intent-based testing? <a href="https://app.daily.dev/feed-by-ids?id=gZJqHPom7&amp;id=j254S16VC&amp;id=sJ5P2kSW8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebMCP will end ‘AI agents clicking buttons’</strong>: Chrome is proposing WebMCP, a new web standard that will allow websites to expose structured tools directly to in-browser AI agents, supposedly ending the era of ‘AI agents clicking buttons.’ This will replace brittle DOM scraping with explicit function calls. So, instead of AI agents clumsily interacting with our UIs, they’ll now have a nice, clean API to exploit. Progress, I guess? <a href="https://app.daily.dev/feed-by-ids?id=Ib422mkez&amp;id=NpNZ22j1b&amp;id=R0k8YNGTx" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
