<h2>Must know</h2>
<ul>
<li><strong>Google Gemini AI faces extraction attempts and malicious use</strong>: Google has blocked over 100,000 attempts to extract proprietary reasoning capabilities from its Gemini AI model, which it considers intellectual property theft. Nation-state actors from China, Iran, North Korea, and Russia are integrating Gemini into attack operations for social engineering and phishing. New malware families like HONESTCUE are embedding Gemini’s API to generate and execute malicious code, bypassing safety filters, as reported by Google’s internal detections and threat report. <a href="https://app.daily.dev/posts/OQLlwLpSw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>TypeScript 6.0 beta released, Go compiler transition confirmed</strong>: TypeScript 6.0 beta is out, marking the final version built on JavaScript before a planned transition to a Go-based compiler for TypeScript 7.0. This move aims for significant performance gains. The beta includes a <code>--stableTypeOrdering</code> flag for migration, support for <code>es2025</code> and new Temporal API types, and improvements to <code>this-less</code> function handling. The production release of 6.0 is expected in March 2026. <a href="https://app.daily.dev/feed-by-ids?id=62kNn8zlZ&amp;id=vVvcWvIPI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.0 kernel updates and post-quantum cryptography integration</strong>: The upcoming Linux 7.0 kernel introduces significant performance optimizations for memory management and XFS file system operations, alongside expanded hardware support for Intel Nova Lake S and ASUS motherboards. Crucially, it integrates ML-DSA, a quantum-resistant signature algorithm, for module signing, replacing SHA-1 support and marking a step towards post-quantum cryptography standards set by NIST. <a href="https://app.daily.dev/feed-by-ids?id=6DJe6t3uU&amp;id=Ukr75F53X&amp;id=Vay0ocYOF&amp;id=Vh5TboDqs&amp;id=m5L4rWAXQ&amp;id=ppwBwOuN5&amp;id=uGdXnDZH4&amp;id=xVcPYGmKq&amp;id=jIFDa9NtZ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Go 1.26 released with green tea GC and new features</strong>: Go 1.26 is out, featuring the Green Tea garbage collector by default, a 30% reduction in <code>cgo</code> call overhead, experimental SIMD operations, and a new <code>errors.AsType</code> for generic error handling. <a href="https://app.daily.dev/feed-by-ids?id=1ObNpwyIZ&amp;id=mj2oUwkea&amp;id=r4TGURizp" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Minimax launches M2.5 AI model at competitive pricing</strong>: Minimax released its M2.5 frontier AI model, priced at $1/hour, claiming it’s over 90% cheaper than Claude Opus or GPT-5, with OpenHands benchmarking it as a top open model for coding tasks. <a href="https://app.daily.dev/posts/VQt2ETdNG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI releases GPT-5.3-Codex-Spark on Cerebras CS3 accelerators</strong>: OpenAI launched GPT-5.3-Codex-Spark, a coding model running on Cerebras CS3 accelerators, achieving over 1,000 tokens per second and prioritizing speed over depth with 15x faster inference. <a href="https://app.daily.dev/feed-by-ids?id=2DKu3o06c&amp;id=2G5oHg0oJ&amp;id=A9s0KznyU&amp;id=CacFMCP2L&amp;id=PFJ429xnc&amp;id=TgVTzT7ZG&amp;id=dqsxZt8l5&amp;id=f8bwmVFOm&amp;id=iVwXrkdSj&amp;id=yYXkpJGbb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>HBM4 memory enters mass production for AI accelerators</strong>: Samsung and Micron have begun mass production and shipping of HBM4 memory, the next-gen high-bandwidth memory for AI accelerators, with Micron’s 2026 output already pre-sold. <a href="https://app.daily.dev/feed-by-ids?id=OUBYQ2PHL&amp;id=WTPyAW1Ca&amp;id=gpkg8awYO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Orchestra and GoClaw enhance self-hosting ecosystem</strong>: Orchestra, an open-source PaaS for bare metal, and GoClaw, an ultra-lightweight Go-based AI assistant for self-hosted OpenClaw, were released, expanding options for managing personal infrastructure and AI agents. <a href="https://app.daily.dev/feed-by-ids?id=0nUdXf1b9&amp;id=C2t5MyKnp&amp;id=jNBhBoH4V&amp;id=nwGZvk6OS&amp;id=sjdf7M4Ia&amp;id=qxYADTM4h" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PVS-Studio to release static analyzer for Go</strong>: PVS-Studio is launching a static analyzer for Go in April 2026, claiming to catch bugs missed by <code>go vet</code> and demonstrating its effectiveness by finding issues in Docker, Kubernetes, and Gitea. <a href="https://app.daily.dev/feed-by-ids?id=1ObNpwyIZ&amp;id=mj2oUwkea" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Server-Sent Events (SSE) often better than WebSockets for real-time apps</strong>: Server-Sent Events (SSE) are highlighted as a simpler and more efficient alternative to WebSockets for 95% of real-time applications that primarily involve server-to-client data flow, offering automatic reconnection and easier integration with existing HTTP infrastructure. <a href="https://app.daily.dev/posts/WPd5fBEqm" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI’s impact on developer roles: junior devs are doomed, seniors are ‘agent bosses’</strong>: Apparently, AI tools are eliminating routine coding tasks, which means junior developers are now obsolete because their ‘grunt work’ is gone. But don’t worry, senior engineers are becoming ‘agent bosses’ managing AI assistants, boosting productivity by 50%. So, if you’re not a senior engineer, maybe start practicing your ‘would you like fries with that’ because the AI overlords have decided your career path. <a href="https://app.daily.dev/feed-by-ids?id=0wFgrbFqP&amp;id=78oOrNVDz&amp;id=B3N2Icy7Y&amp;id=BKJ8NHw6d&amp;id=DcSKM9BQA&amp;id=GManH4sq3&amp;id=GnwSkM7qk&amp;id=HHMxTl4yN&amp;id=JgHE2Ywe8&amp;id=MPMZLgvp3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI hype, governance, and economic realities: 95% failure rate, but don’t worry, it’s a bubble!</strong>: AI companies are setting ‘impossibly high expectations’ with hyperbolic claims, leading to a 95% failure rate in generating measurable value for corporations. But hey, Sam Altman predicts GPT-5 will be 2x smarter and 100x cheaper by late 2027, so just ignore the current reality where AI-generated code has 322% more privilege-escalation risks. It’s fine, some experts say the ‘AI bubble’ bursting will be ‘beneficial,’ leaving behind cheap GPUs. What a relief! <a href="https://app.daily.dev/feed-by-ids?id=1ssjvObXV&amp;id=6C3DC1GxE&amp;id=7qf6CcnTl&amp;id=KBiI9wOrx&amp;id=NO2cmpN4i&amp;id=PTlBRJtuz&amp;id=PryOnuBPQ&amp;id=Sfm1OU6it&amp;id=ZJJH0MAGc&amp;id=mcpWOltJw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes is overkill for 99% of apps, says team running 500k logs/day on Docker Compose</strong>: A team proudly declares Kubernetes is ‘overkill for 99% of apps’ because they’re handling 500k logs/day on a single server with Docker Compose for $14/month. They even use PostgreSQL for everything, including queues and pub/sub. So, all you folks wrestling with Kubernetes complexity, apparently you’re just doing it wrong. Just use Docker Compose and a single Postgres instance, it’s that simple. Who needs distributed systems anyway? <a href="https://app.daily.dev/posts/PNwYwcQ5e" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
