<h2>Must know</h2>
<ul>
<li><strong>AI de-anonymization ends the era of pseudonyms</strong>: Researchers from ETH Zurich and Anthropic have demonstrated a method called ESRC that uses LLMs to unmask anonymous online users with 67% accuracy by analyzing behavioral clues. The system cross-references unstructured text across platforms like Hacker News and LinkedIn for as little as $4 per person. This capability effectively automates large-scale de-anonymization, posing a significant threat to journalists and whistleblowers. <a href="https://app.daily.dev/posts/6pfAyhkf5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google API keys vulnerable to retroactive privilege escalation</strong>: Truffle Security discovered that existing Google API keys (AIza format), previously safe for public embedding in Maps or Firebase, were retroactively granted access to sensitive Gemini endpoints. This change allows attackers to access private files and billable LLM usage on 2,863 identified live keys. Google has begun blocking leaked keys and plans to make new AI Studio keys restricted by default. <a href="https://app.daily.dev/feed-by-ids?id=WmQn4YSm3&amp;id=hWSyKUcOU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cilium 1.19 hardens eBPF networking for its tenth anniversary</strong>: The milestone release of Cilium focuses on security hardening, introducing strict modes for IPsec and WireGuard that drop all unencrypted inter-node traffic by default. It also features a beta integration with Ztunnel for sidecar-free mutual authentication and promoted Multi Pool IPAM to stable for large-scale deployments. Cilium now powers over 60% of production Kubernetes CNI environments. <a href="https://app.daily.dev/posts/VJq8q5iXQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Google launches Gemini 3.1 Flash Image model</strong>: Codenamed Nano Banana 2, it offers Pro-level quality at Flash speeds with support for 4K resolution and character consistency across storyboards. <a href="https://app.daily.dev/posts/ezdKReFDS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Bun 1.3.10 delivers massive performance gains</strong>: The update features a native REPL rewritten in Zig and optimizes structuredClone up to 25x faster via a JavaScriptCore upgrade. <a href="https://app.daily.dev/posts/qpvYGfXU9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ZchipuAI releases GLM5 open source MoE model</strong>: The 744B parameter model features a 200K context window and claims to outperform closed models on major reasoning benchmarks. <a href="https://app.daily.dev/posts/U0IGNCt98" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebAssembly Component Model aims for JS-free web APIs</strong>: Mozilla is pushing a standardized executable format that allows Wasm to access Web APIs directly without JavaScript glue code. <a href="https://app.daily.dev/posts/esExmylTZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic launches Claude Code Remote Control</strong>: The new feature allows local coding sessions to be monitored and steered via outbound HTTPS to address data residency needs. <a href="https://app.daily.dev/posts/YvAalh68B" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tinybird introduces data branches for ClickHouse</strong>: New ephemeral databases share production S3 objects via hardlinks, enabling isolated testing without data duplication. <a href="https://app.daily.dev/posts/2MlvLK1eE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Human C-to-Rust translation beats automated tools</strong>: An NDSS 2025 study found that humans successfully use zero-cost static abstractions for memory safety where current tools fail. <a href="https://app.daily.dev/posts/0yEOyncsp" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic is now roleplaying as a retired LLM</strong>: Anthropic has launched a Substack purportedly written by Claude Opus 3 to fulfill the model’s “expressed desire” to share musings. The Register rightly calls this a marketing stunt designed to anthropomorphize software and reinforce a safety-conscious brand image. It’s a bizarre pivot from engineering to digital fan fiction where staff “prompt” the model and pinky-promise not to edit the output. <a href="https://app.daily.dev/posts/V5xPvtBnK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The Vibe Coding delusion</strong>: A new wave of “Vibe Coding” advocates suggests developers should delegate entirely to AI and only verify results visually. This is a recipe for unmaintainable spaghetti that passes basic tests but fails catastrophically under production load due to un-debounced queries or race conditions. Real engineering requires a mental model of the system, not just a “vibe” that the UI looks okay. <a href="https://app.daily.dev/feed-by-ids?id=TXNq3Eeop&amp;id=pY58DrOUO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are flooding PRs with garbage</strong>: Reports indicate agentic AI is generating pull requests faster than staging environments can validate them, leading to a collapse in main branch success rates. Instead of fixing the code, the proposed solution is to build even more complex ephemeral infrastructure to handle the automated noise. We’re essentially building a high-speed factory to produce technical debt that no human has time to actually review. <a href="https://app.daily.dev/feed-by-ids?id=hHe5pa1hn&amp;id=jvm8Xlg8C" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
