<h2>Must know</h2>
<ul>
<li><strong>Meta and Google double down on AI infrastructure and multimodal models</strong>: Meta is launching ‘Meta Compute,’ a massive infrastructure scale-up aiming for tens of gigawatts of power capacity to support future AI workloads, as reported by their own executive announcements. This puts them in direct competition with Microsoft and Google for generative AI cloud dominance. Meanwhile, Google released Gemini 3, a multimodal AI platform processing diverse data types with a 1M+ token context window, and Veo 3.1, which now produces native vertical videos for social platforms. These moves signal an escalating arms race in foundational AI capabilities and infrastructure. <a href="https://app.daily.dev/feed-by-ids?id=2S31ZjKRT&amp;id=NEGZugNBg&amp;id=kwRMqebwB&amp;id=qjYQj2ENJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s token crackdown sparks developer backlash and open-source alternatives</strong>: Anthropic has begun strictly enforcing its terms, preventing the use of Claude Code subscription tokens with third-party tools like Cursor and Open Code, forcing users to pay significantly higher API prices for external access. This move, while technically within their rights, has caused considerable controversy among developers who built workflows around these integrations, as noted in technical reporting. The timing coincides with the rise of open-source alternatives like grepai, which offers intent-based semantic search with local vector embeddings, eliminating reliance on cloud infrastructure and offering increased token efficiency, directly challenging Anthropic’s closed ecosystem approach. <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>AI agents face critical security and governance challenges</strong>: Agentic AI systems, while promising, are proving to be unreliable and insecure by design, according to Signal president Meredith Whittaker. Examples like Microsoft’s Recall highlight vulnerabilities to malware and prompt injection attacks, as user activity is stored in databases. Gartner predicts 40% of agentic AI projects will fail by 2027 due to inadequate guardrails, emphasizing the need for strict governance, technical constraints, and phased rollouts. This underscores the critical importance of ethical considerations and robust security in AI development, especially as these systems gain more autonomy. <a href="https://app.daily.dev/feed-by-ids?id=TJt6uaOsV&amp;id=XnmeLgefE&amp;id=rjeAcjlYe&amp;id=9X1ifJmCu&amp;id=QnCK3ryB4" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Open-weight AI models challenge proprietary leaders in coding tasks</strong>: New open-weight AI models like GLM 4.7 and MiniMax M2.1 are demonstrating near-frontier capabilities in complex autonomous coding tasks, with GLM 4.7 completing a 20-feature CLI task runner in 10 minutes and MiniMax M2.1 achieving similar results at half the cost, showcasing significant progress in open-source AI. <a href="https://app.daily.dev/feed-by-ids?id=p3khbP8jU&amp;id=sDf9ENrg4&amp;id=7kuJfwmkk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WordPress 7.0 to drop older PHP support</strong>: The WordPress core team is preparing for version 7.0, which will discontinue support for PHP 7.2 and 7.3, aligning with ongoing platform updates and potentially impacting developers and hosting providers. <a href="https://app.daily.dev/posts/xho5u0zys" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux Mint 22.3 enhances accessibility and graphical tools</strong>: Linux Mint 22.3 ‘Zena’ introduces a redesigned Mint Menu, an on-screen keyboard for improved accessibility, and enhanced file manager features, all built on Ubuntu 24.04.3 LTS with a focus on expanding GUI-based administration. <a href="https://app.daily.dev/posts/ecK1ViBJk" 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) have been released with security updates addressing multiple CVEs, focusing on permission handling, error management, and updated external dependencies. <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>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>Python Software Foundation receives $1.5M for security</strong>: Anthropic has pledged $1.5 million over two years to the Python Software Foundation (PSF) to enhance Python ecosystem security, focusing on automated malware detection for PyPI packages and supporting CPython core development. <a href="https://app.daily.dev/posts/yCHi0TB7q" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI shifts ecology research and biodiversity investment</strong>: Ecologists are increasingly using AI, machine learning, and automated sensors for data collection, while a new $25.9 million fund from Superorganism targets biodiversity startups, signaling a technological shift in conservation efforts. <a href="https://app.daily.dev/feed-by-ids?id=YsLIINgTC&amp;id=dHxdm7wPU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Cloudflare fined for not playing piracy police</strong>: Italy’s AGCOM fined Cloudflare €14.2 million for refusing to block pirate sites on its 1.1.1.1 DNS resolver, citing technical infeasibility and potential latency for all users. Cloudflare CEO Matthew Prince is now reaching out to the U.S. government and even the IOC, threatening to pull servers from Italy. Because nothing says ‘internet freedom’ like a company threatening to disrupt services over a copyright dispute, all while claiming it’s ‘impossible’ to comply without broader negative impacts. Convenient. <a href="https://app.daily.dev/posts/Q0IbcQd4P" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft’s ‘Community-First AI’ initiative: greenwashing or genuine change?</strong>: Microsoft announced a ‘Community-First AI Infrastructure’ initiative, promising to cover power costs, minimize water usage, and create local jobs for its U.S. data centers, directly responding to widespread public backlash. They’re even foregoing local tax breaks to prevent residential electricity bill hikes. It’s almost as if they realized that building massive, resource-hungry AI data centers without community consent might be a bad look. Let’s see if these promises actually stick, or if it’s just a PR-friendly way to keep expanding their energy footprint. <a href="https://app.daily.dev/posts/jqtBl3Jnb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI code generation: the verification bottleneck is the new toil</strong>: Apparently, AI generating code faster doesn’t mean developers work less; it just means they spend more time verifying the AI’s unreliable output. The New Stack reports that 96% of developers don’t fully trust AI-generated code, leading to increased technical debt and correction work. So, we’ve successfully shifted ‘toil’ from writing code to babysitting AI. Progress! <a href="https://app.daily.dev/posts/SVMT3qR0j" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
