<h2>Must know</h2>
<ul>
<li><strong>AI agent security vulnerabilities escalate</strong>: New research from Snyk reveals critical security flaws in the AI Agent Skills ecosystem, with 13.4% of scanned skills containing critical issues and 76 confirmed malicious payloads designed for credential theft and data exfiltration. This highlights a significant supply chain risk for enterprises adopting AI agents. Operant AI launched Agent Protector, a real-time security solution for autonomous AI agents, offering inline protection against privilege escalation and data exfiltration, emphasizing the urgent need for robust security in agentic systems. <a href="https://app.daily.dev/feed-by-ids?id=0BdgmYNfg&amp;id=0e28QTlYb&amp;id=2vNCw5dB4&amp;id=dL9ys7HAk&amp;id=the2EenMO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic and OpenAI clash over AI monetization strategies</strong>: Anthropic plans Super Bowl ads mocking OpenAI’s move to introduce advertising in ChatGPT, positioning its Claude AI as ad-free. This public spat highlights differing business models, with OpenAI reportedly burning $9 billion annually and Anthropic relying on enterprise contracts. Meanwhile, OpenAI launched ‘Frontier,’ an enterprise platform for building and managing AI agents, while Anthropic released Claude Opus 4.6 with expanded context windows and ‘agent teams,’ further intensifying competition in the enterprise AI space. <a href="https://app.daily.dev/feed-by-ids?id=TmM2mNSSP&amp;id=TGZyay3jK&amp;id=G7Sre82A1&amp;id=Ijw4mCG7B" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI bot traffic threatens web publishers</strong>: AI bots, primarily retrieval augmented generation (RAG) bots, now account for a significant portion of web traffic, reaching one bot visit for every 31 human visits by Q4 2025. This surge has caused clickthrough rates from AI applications to source websites to plummet to 0.27%, severely impacting publishers’ financial viability. Publishers are engaging in an ‘arms race’ against increasingly sophisticated AI bots, fundamentally reshaping how the internet functions and who primarily uses it. <a href="https://app.daily.dev/posts/5O2tzW5oh" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Microsoft appoints engineering quality czar</strong>: Microsoft CEO Satya Nadella appointed Charlie Bell as engineering quality czar to address ongoing issues like Azure outages and Windows patch problems, with Hayete Gallot returning to lead security. <a href="https://app.daily.dev/posts/NGytU6zYE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS launches new EC2 instances</strong>: AWS introduced new EC2 High Memory U7i-6TB instances and C8id, M8id, and R8id instances powered by custom Intel Xeon processors, offering significant performance and memory bandwidth improvements for various workloads. <a href="https://app.daily.dev/posts/JPt4mjvME" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ScyllaDB integrates USearch for vector search</strong>: ScyllaDB now includes vector search capabilities using the open-source USearch library, allowing storage of embeddings alongside structured data and leveraging its shard-per-core architecture for optimized performance in real-time AI applications. <a href="https://app.daily.dev/posts/pKNE7iWlw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Laravel releases AI SDK for application development</strong>: Laravel launched an official AI SDK in beta, providing a unified API for integrating AI capabilities from multiple providers like Anthropic, Gemini, and OpenAI into Laravel applications, supporting features like agents, memory, and structured outputs. <a href="https://app.daily.dev/posts/ND9RAWZ9f" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Intel Arc B390 Linux performance analyzed</strong>: Intel Arc B390 graphics with Xe3 architecture show competitive performance on Linux when properly configured, but out-of-the-box experience is hampered by lower default power limits compared to Windows. <a href="https://app.daily.dev/posts/GxsXcKIKI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GNOME Shell and Mutter 50 beta released</strong>: The GNOME Shell and Mutter 50 beta releases introduce stable Variable Refresh Rate (VRR) support, improved frame scheduling, and enhanced discrete GPU detection, paving the way for the stable GNOME 50 release in March. <a href="https://app.daily.dev/posts/buHOzLdcP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust ecosystem update #637</strong>: The latest ‘This Week in Rust’ newsletter highlights new library additions, stabilized features like <code>if let</code> guards, improved compiler performance, and community articles on network programming and SIMD benchmarking. <a href="https://app.daily.dev/posts/HbfZ0FSX1" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Elixir is the ‘best language for AI’</strong>: An article boldly claims Elixir is the ‘best language for AI,’ citing an unverified ‘Tencent study’ that supposedly found Elixir achieved a 97.5% LLM completion rate. Because nothing says ‘unbiased technical assessment’ like a single, unsourced study from a company that definitely isn’t trying to push its own tech. Sure, Elixir’s great, but ‘best for AI’ based on a phantom study? Give me a break. <a href="https://app.daily.dev/posts/1D7tGyRwk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>UK government’s deepfake detection plan</strong>: The UK government, in its infinite wisdom, announced plans to develop a framework for evaluating deepfake detection technologies, partnering with Microsoft and academics. Cybersecurity expert Dr. Ilia Kolochenko, however, points out that existing open-source tools already do this, and the real problem is enforcement, not detection. So, another government initiative to reinvent the wheel while ignoring the actual steering problem. Brilliant. <a href="https://app.daily.dev/posts/EQCYBDiDB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI will make software slower than hardware</strong>: A new article warns that LLMs might finally break Wirth’s Law, making software slower faster than hardware gets faster, because apparently, repeatedly querying expensive models for simple tasks is the pinnacle of efficiency. The author suggests LLMs will erode fundamental programming skills, leading to a ‘thought-terminating approach’ to problem-solving. Because, you know, we were all so efficient before AI, and nobody ever wrote bad code that ran slowly. <a href="https://app.daily.dev/posts/SELxucgnM" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
