<h2>Must know</h2>
<ul>
<li><strong>Anthropic tightens Claude Code token use, developers revolt</strong>: Anthropic is now strictly enforcing its terms of service, preventing users from leveraging Claude Code subscription tokens with third-party tools like Cursor and Open Code. This move, which requires users to either use Claude Code exclusively or pay significantly higher API prices for external integrations, has sparked considerable backlash among developers who had built workflows around these integrations. The company cites debugging challenges and unsustainable infrastructure costs, while developers feel betrayed by the abrupt change after months of leniency, as reported by multiple tech outlets. <a href="https://app.daily.dev/feed-by-ids?id=DEqFrm707&amp;id=jcpFkjQkx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Meta pivots from metaverse to AI, lays off 10% of Reality Labs</strong>: Meta is closing three VR studios (Twisted Pixel Games, Sanzaru Games, Armature Studio) and laying off approximately 1,500 Reality Labs staff, about 10% of the division. This strategic shift comes after Reality Labs lost over $70 billion since 2021, with Meta now redirecting investment towards AI-powered glasses and wearable technology. The company is also scaling up its AI infrastructure with Meta Compute, aiming for massive power capacity to support future AI workloads, according to Meta’s official announcements and industry reports. <a href="https://app.daily.dev/feed-by-ids?id=7ystcAQPc&amp;id=2S31ZjKRT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare fined by Italy for not blocking pirate sites</strong>: Italy’s communications regulator, AGCOM, has fined Cloudflare €14.2 million for refusing to block pirate sites on its 1.1.1.1 DNS resolver, citing the Piracy Shield law. Cloudflare argues that compliance is technically infeasible without increasing latency for all users and potentially blocking legitimate sites, stating ‘Non C’è Modo’ (No Way). CEO Matthew Prince is seeking U.S. government support and has warned the IOC about potential cybersecurity risks for the 2026 Winter Olympics in Italy, threatening to withdraw servers from the country if forced to comply, according to Cloudflare’s public statements. <a href="https://app.daily.dev/posts/Q0IbcQd4P" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Open-weight AI models match frontier capabilities at lower cost</strong>: New open-weight AI models, GLM 4.7 and MiniMax M2.1, demonstrated the ability to autonomously build complex CLI task runners, with GLM 4.7 completing the task in 10 minutes and MiniMax M2.1 at half the cost, showcasing agentic behavior and closing the gap with proprietary models. <a href="https://app.daily.dev/posts/p3khbP8jU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New Linux malware targets cloud and containers</strong>: Check Point Research identified VoidLink, a new Linux malware framework written in Zig, designed to infiltrate cloud and container environments (AWS, Azure, GCP, Docker, Kubernetes) using advanced rootkit techniques and a modular plugin system for credential harvesting and lateral movement. <a href="https://app.daily.dev/posts/uJaIfhgHH" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LangSmith launches no-code AI agent builder</strong>: LangSmith has released its Agent Builder, allowing users to create AI agents for complex, multi-step tasks using natural language instructions, without coding, and supporting team sharing, custom tools, and API embedding. <a href="https://app.daily.dev/posts/sDf9ENrg4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WordPress 7.0 to drop PHP 7.2 and 7.3 support</strong>: The WordPress core development team plans to discontinue support for PHP 7.2 and 7.3 in the upcoming WordPress 7.0 release, with a meeting scheduled to finalize volunteer participation and other development-focused agenda items. <a href="https://app.daily.dev/posts/xho5u0zys" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js releases security updates</strong>: Node.js 20.20.0 (LTS) and 25.3.0 (Current) were released with security updates addressing multiple CVEs, enhancing permission handling, and improving error management for TLSSocket and async_hooks. <a href="https://app.daily.dev/feed-by-ids?id=DwsoMgMKH&amp;id=Wy4br4Zyw&amp;id=Zr0Dv9dwZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Bun 1.3.6 adds tarball handling and performance boosts</strong>: Bun v1.3.6 introduces Bun.Archive for tarball creation/extraction, a JSONC parser, and significant performance improvements, including 3.5x faster Response.json() and 20x faster Bun.hash.crc32, alongside 45 bug fixes. <a href="https://app.daily.dev/feed-by-ids?id=Wx4br4Zyw&amp;id=Yy5NFIIOM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux Mint 22.3 enhances accessibility and GUI tools</strong>: Linux Mint 22.3 “Zena” features a redesigned Mint Menu, a new on-screen keyboard for improved accessibility, enhanced language input, and new graphical tools for system hardware info and GRUB editing, all based on Ubuntu 24.04.3 LTS. <a href="https://app.daily.dev/posts/ecK1ViBJk" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents are unreliable and insecure, says Signal</strong>: Signal’s Meredith Whittaker and VP are out here saying agentic AI systems, like Microsoft’s Recall, are ‘unreliable and insecure by design.’ Apparently, these probabilistic wonders fail at every step, making multi-step tasks a total crapshoot. They’re calling for a halt to ‘reckless AI deployment’ and demanding transparency, because who needs a chatbot that agrees to sell a $76,000 car for $1, right? It’s almost like these things aren’t magic. <a href="https://app.daily.dev/feed-by-ids?id=TJt6uaOsV&amp;id=XnmeLgefE&amp;id=rjeAcjlYe&amp;id=QnCK3ryB4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft’s ‘Community-First AI Infrastructure’ initiative: because greenwashing is cheaper than actual green energy</strong>: Microsoft, facing backlash from 142 activist groups across 24 states, has announced its ‘Community-First AI Infrastructure’ initiative, promising to fully cover power costs for its U.S. data centers and minimize water usage. This comes after abandoning projects in Wisconsin and facing resistance in Michigan and Ohio. So, instead of actually building sustainable infrastructure from the get-go, they’re now promising to be good neighbors after the fact. How very ‘community-first’ of them to only care when the pitchforks come out. <a href="https://app.daily.dev/posts/jqtBl3Jnb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is killing open source documentation, but don’t you dare try to fix it</strong>: Tailwind CSS is reportedly seeing an 80% revenue drop because AI is just scraping their documentation, and they’re so mad about it they rejected a pull request to improve LLM parsing. Because, you know, the solution to AI eating your business model is to make your documentation harder for AI to read, not to adapt. Meanwhile, open-source maintainers are switching to AGPL-3.0 to stop AI from training on their code. It’s a real ‘damned if you do, damned if you don’t’ situation, unless you’re a cloud service, then you’re raking it in. <a href="https://app.daily.dev/feed-by-ids?id=2a8HRxzCb&amp;id=Ml47Y2Hd6&amp;id=g50WJmopi" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
