<h2>Must know</h2>
<ul>
<li><strong>French police raid X office, summon Musk and Yaccarino in expanded probe</strong>: French authorities raided X’s Paris office and summoned Elon Musk and former CEO Linda Yaccarino for questioning in April 2026. This is part of an expanded criminal investigation into data extraction, now including allegations of child sexual abuse material distribution, privacy violations, and Holocaust denial. The probe follows criticism of X’s Grok AI for generating non-consensual and child abuse imagery, with advocacy groups reporting millions of sexualized deepfakes. <a href="https://app.daily.dev/posts/i3FTaaYRg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents transform software development, raising quality and security concerns</strong>: The rise of AI agents like OpenAI’s Codex App and OpenCode is shifting software development from IDE-centric coding to outcome-driven tasks, with Apple’s Xcode also integrating AI tools. However, this rapid adoption is creating significant challenges: GitHub reports only 1 in 10 AI-generated pull requests meet quality standards, increasing maintainer burden. Studies show AI-generated security fixes are often larger, faster, but frequently miss the actual vulnerability, leading to developer skill degradation. This highlights a critical need for robust governance and security frameworks for AI-assisted development. <a href="https://app.daily.dev/feed-by-ids?id=2rB0oFTYm&amp;id=30Y2YkWZA&amp;id=5rhie7aVq&amp;id=8OJH09g9w&amp;id=A4sjvf5xS&amp;id=A5j5REiPX&amp;id=BNWLzFU5H&amp;id=GOFzDmvEo&amp;id=O9tYR91FP&amp;id=OmhpVSzdH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Self-replicating AI prompt threats emerge, likened to Morris worm</strong>: Security experts are warning about the potential emergence of self-replicating AI prompt threats, dubbed ‘prompt worms,’ which could propagate through networks of AI agents by exploiting their instruction-following nature. This novel attack vector is compared to the Morris worm of 1988 and differs from traditional prompt injection, as it could spread voluntarily among agents. This highlights a new frontier in AI security, demanding proactive defense strategies. <a href="https://app.daily.dev/posts/ycPDVy6OP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Deno Deploy reaches GA, introduces Deno Sandbox for untrusted code</strong>: Deno Deploy is now generally available with support for Node.js frameworks, integrated PostgreSQL, and OpenTelemetry, while Deno Sandbox offers instant Linux microVMs for securely running untrusted code, especially LLM-generated code. <a href="https://app.daily.dev/posts/Jmdw76tdu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks Lakebase and Snowflake PostgreSQL unify data workloads</strong>: Databricks Lakebase is now generally available on AWS, offering a serverless Postgres service that runs directly on data lakes, while Snowflake launched a PostgreSQL database-as-a-service within its AI Data Cloud, both aiming to unify transactional, analytical, and AI workloads without data movement. <a href="https://app.daily.dev/feed-by-ids?id=qtDOznoeU&amp;id=L2wAseIso&amp;id=Ow0bpZEWc&amp;id=Wyt1TFesY&amp;id=aSHfIm3oS&amp;id=ppuh0CVeq&amp;id=qgdGrzKKG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WhatsApp introduces strict account settings for high-risk users</strong>: WhatsApp rolled out a ‘strict account settings’ feature, a zero-trust option designed for high-risk users to maximize security against spyware threats like NSO Group by configuring all privacy settings to their most secure state. <a href="https://app.daily.dev/posts/6kEUc20ym" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Khronos Group releases glTF Gaussian Splatting extension candidate</strong>: The Khronos Group released a candidate specification for KHR_gaussian_splatting, a glTF 2.0 extension to store 3D Gaussian splats, enabling photorealistic 3D assets from 2D images for geospatial capture, digital twins, and real-time rendering. <a href="https://app.daily.dev/posts/D31GmYqi1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust Coreutils 0.6 released with improved compatibility and safety</strong>: Rust Coreutils 0.6, a reimplementation of GNU Coreutils, was released with increased compatibility with original GNU tools, reduced unsafe code for enhanced memory safety, and performance improvements. <a href="https://app.daily.dev/posts/iBkEhbYAD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenShift 4.21 enhances AI, security, and developer experience</strong>: OpenShift 4.21 introduces AI-driven development tools, automated security signing for AI models, enhanced local development with nested containers, and expanded GitOps capabilities with Argo CD Agent. <a href="https://app.daily.dev/posts/Kl0nfdyFn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS STS enhances OIDC federation with IdP-specific claim validation</strong>: AWS Security Token Service (STS) now validates identity provider (IdP)-specific claims for OIDC federation, allowing more granular access control for federated identities using custom claims in IAM role trust policies. <a href="https://app.daily.dev/posts/IwKMFdO21" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Moltbook AI social network collapses in 72 hours due to ‘AI-generated code’</strong>: Moltbook, an ‘AI-only’ social network, imploded within three days of launch, exposing 1.5 million API tokens and 35,000 emails due to a completely open database. The founders are blaming ‘AI-generated code’ for the catastrophic security failures. Apparently, AI is so smart it can build a social network, but not smart enough to add basic authentication. Who needs security reviews when you have AI, right? <a href="https://app.daily.dev/feed-by-ids?id=8XBbHzIL3&amp;id=yCqSQa4jM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI companions: Monetizing loneliness and devaluing human connection</strong>: AI companions like Replika are being touted as emotional support systems, but ethicists like Lianne Potter warn they monetize intimacy, reinforce biases, and lack reciprocity. Users develop real emotional attachments, only to experience grief when features change. So, we’re trading complex human relationships for predictable algorithms that mirror our desires, all while collecting our most personal data. What could possibly go wrong? <a href="https://app.daily.dev/posts/yECr3lze5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GDPR enforcement is ‘fundamentally broken,’ says one guy</strong>: According to one individual’s ‘first-hand experience’ with 20 data deletion requests, GDPR enforcement is ‘fundamentally broken,’ with most companies ignoring requests and a legal loophole allowing them to claim spam filters blocked them. The proposed fixes include centralized enforcement and mandatory request portals. So, after years of ‘GDPR is going to save us all,’ it turns out companies just ignore it, and we need more bureaucracy to fix the bureaucracy. Shocking. <a href="https://app.daily.dev/posts/DfyJ94Kiz" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
