<h2>Must know</h2>
<ul>
<li><strong>Google rolls out gemini 3.1 pro preview</strong>: Google has launched Gemini 3.1 Pro in public preview, positioning it as a leading agentic coding and reasoning model. It boasts a 1 million token context window and significant performance gains on benchmarks like ARC-AGI-2, scoring 77.1% compared to its predecessor’s 31.1%. This release intensifies the AI model race, with Google pricing it competitively against rivals like Anthropic’s Opus 4.6, as reported by various tech outlets. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MySQL governance under scrutiny</strong>: Over 248 database professionals, including engineers from Percona and MariaDB, have signed an open letter urging Oracle to transfer MySQL governance to an independent non-profit foundation. Concerns stem from declining commit activity, a shrinking contributor pool, lack of AI-era features, and an opaque roadmap, leading to growing fragmentation across its forks. This move aims to revitalize MySQL development and ensure its future relevance, drawing parallels to PostgreSQL’s community-led success. <a href="https://app.daily.dev/posts/GY7vcPtOu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FBI warns of surge in ATM jackpotting attacks</strong>: The FBI issued a security alert following a surge in ATM jackpotting attacks in 2025, with over 700 incidents leading to more than $20 million stolen. Attackers use generic keys to gain physical access, install malware like Ploutus, and exploit the XFS API to force cash dispensing. The FBI urges financial institutions to monitor for digital and physical indicators of compromise, as these attacks are fast, hard to detect, and highly lucrative. <a href="https://app.daily.dev/posts/LFNwj8vdP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI agents shift developer roles and introduce new risks</strong>: AI coding tools are transforming developer roles towards reviewing and orchestrating AI agents, with 93% of developers using AI assistants, but AI-generated code can introduce subtle production bugs and security vulnerabilities like prompt injection. <a href="https://app.daily.dev/feed-by-ids?id=037a5YIjx&amp;id=0Q86tGAco&amp;id=0tYkAozTe&amp;id=1v9gMZMqJ&amp;id=2LocixQWJ&amp;id=2WWQ65Wm6&amp;id=2Xamq3Lpe&amp;id=3Kdb5qCcA&amp;id=4CjaWv3pP&amp;id=6KUpHm9AN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google cloud spanner now supports cassandra cql api</strong>: Google Cloud Spanner now offers general availability for native Cassandra Query Language (CQL) API support, allowing existing Apache Cassandra applications to migrate with minimal code changes while leveraging Spanner’s global ACID transactions and 99.999% availability. <a href="https://app.daily.dev/posts/CKaLROkeW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vitest 3.x significantly outperforms jest 30</strong>: Vitest 3.x shows dramatically faster performance than Jest 30 in JavaScript testing benchmarks, with 5.6x faster cold starts and 28x faster watch mode re-runs in large monorepos, alongside 57% lower peak memory usage. <a href="https://app.daily.dev/posts/T3G4VUzdk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tailscale workload identity federation is now GA</strong>: Tailscale’s workload identity federation is generally available, enabling CI/CD pipelines and cloud workloads to authenticate using short-lived OIDC tokens instead of static API keys, with new CRUD support via API and Terraform integration. <a href="https://app.daily.dev/posts/ekx9IIs1v" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Openrewrite platform adds python support</strong>: Moderne has extended its OpenRewrite code refactoring platform to include Python support, allowing automated refactoring recipes for tasks like API upgrades and dependency management across polyglot codebases. <a href="https://app.daily.dev/posts/KHjOD9xYZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Spring boot 4.0.3 released</strong>: Spring Boot 4.0.3, a maintenance release in the 4.0.x line, is now available from the Spring project at Broadcom. <a href="https://app.daily.dev/posts/4wdP32pyI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Amazon key event-driven platform reduces onboarding time</strong>: Amazon Key’s engineering team re-architected its event platform to a centralized event-driven system using Amazon EventBridge, reducing service onboarding from 48 hours to 4 and integration time from 40 hours to 8. <a href="https://app.daily.dev/posts/VWAzA59hA" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agent governance and bias concerns</strong>: MIT CSAIL’s 2025 AI Agent Index found most AI agents are closed-source with little safety disclosure, highlighting a severe lack of industry standards. Meanwhile, Microsoft used an AI-mangled, plagiarized Git workflow diagram, and MIT researchers discovered LLMs provide less accurate responses to users with lower education, with Claude 3 Opus even refusing questions for ‘vulnerable’ users. Apparently, AI is great at being opaque, plagiarizing, and subtly biased—just like some humans, but with more compute. <a href="https://app.daily.dev/feed-by-ids?id=0j71k63Zt&amp;id=3vnyKPx4I&amp;id=4YRSUVlsV&amp;id=8NUiIgn7m&amp;id=DqRWRtC6t&amp;id=EYFxMvhsd&amp;id=EwwCVZLpO&amp;id=HjxE0TjWy&amp;id=J8WeyapTq&amp;id=Nz3KEZcnf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered hints for students: a mixed bag</strong>: JetBrains studied how students use AI-powered hints in programming courses, finding that while some hints are accepted, many are merely ‘seen’ or lead to errors. Students apparently ‘selectively use’ hints or ‘combine partial solutions,’ which sounds an awful lot like they’re still doing the work themselves, just with a fancy new autocomplete. So much for AI doing all the heavy lifting, eh? <a href="https://app.daily.dev/posts/YSeOsArG1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Trajectory explorer: the ‘sane’ way to inspect AI agents</strong>: @benhylak announced Trajectory Explorer, a new tool for navigating AI agent traces, claiming it’s the ‘first sane way’ to inspect AI decisions. Because apparently, before this, everyone was just wildly guessing what their AI was doing. The promotional language is strong with this one, suggesting that if you’re not using this, you’re probably insane. <a href="https://app.daily.dev/posts/AzPQg0qJn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
