<h2>Must know</h2>
<ul>
<li><strong>AI coding tools boost throughput but degrade code quality</strong>: A Carnegie Mellon University study of 807 GitHub repositories found that while AI coding assistants initially increased code generation by 30-40%, this productivity gain vanished after three months. Worse, static analysis warnings rose by 30% and code complexity by over 40%, persisting long-term. This aligns with 2024 data from GitClear and suggests that while tools like Cursor can accelerate development, they introduce significant technical debt, potentially worsening future AI model training outcomes if not addressed. Meanwhile, Q1 2026 benchmarks across 219 companies show AI coding tools like Cursor and Claude increasing pull request throughput, with GitHub Copilot reaching 9.76% daily adoption, but concerns about code quality persist. <a href="https://app.daily.dev/feed-by-ids?id=EZUXhYwRZ&amp;id=XkFn8TQqF&amp;id=7KqSlBkW9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical vulnerabilities in Chainlit AI framework allow server exploitation</strong>: Two critical vulnerabilities (CVE-2026-22218, CVE-2026-22219) in Chainlit, a popular Python AI chatbot framework with over 5 million downloads, allowed unauthenticated attackers to read any file on the server and conduct CSRF/SSRF attacks. This could lead to the exfiltration of API keys, credentials, and database contents, enabling lateral movement within enterprise networks. Patches are available in Chainlit version 2.9.4, urging immediate upgrades for affected users. <a href="https://app.daily.dev/posts/I9vckVFsM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI addresses Stargate datacenter concerns amid local pushback</strong>: OpenAI is publicly committing to address local concerns regarding electricity costs, water consumption, and environmental impact for its $500 billion Stargate datacenter project. This comes as 20 US datacenter projects were blocked or delayed in Q2 2025 due to community resistance. OpenAI pledges self-funded energy infrastructure and closed-loop cooling systems, but critics remain skeptical about the limited job creation (around 100 full-time roles per campus) and the true environmental mitigation, especially given the lack of independent verification or specific location details. <a href="https://app.daily.dev/posts/O0VoWxWDb" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>LLVM bans unchecked AI-assisted code contributions</strong>: The LLVM compiler project has implemented a ‘human in the loop’ policy, banning AI-generated code contributions without thorough human review, citing a surge in low-quality submissions degrading project quality. This mirrors cURL’s decision to end its bug bounty program due to AI-generated spam. <a href="https://app.daily.dev/feed-by-ids?id=DXhDzQ15v&amp;id=zNsgAYHqy&amp;id=ssDvZHbZq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Amazon launches Bedrock AgentCore for AI agent deployment</strong>: Amazon introduced Bedrock AgentCore, a platform for deploying and operating AI agents with zero infrastructure management, offering modular services like Runtime, Memory, Gateway, Code Interpreter, and Observability, along with a CLI toolkit for rapid creation and integration with frameworks like LangGraph and CrewAI. <a href="https://app.daily.dev/posts/PbCI8QH1q" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>BlackIce toolkit targets AI security testing</strong>: BlackIce is an open-source Dockerized toolkit bundling 14 AI security testing tools, including prompt injection, jailbreak detection, and supply chain scanning, mapped to MITRE ATLAS and DASF frameworks, aiming to simplify AI red teaming and security validation. <a href="https://app.daily.dev/posts/PSfiaCphA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mesa 26.0-rc1 enhances AMD, Intel, and NVIDIA GPU drivers</strong>: Mesa 26.0-rc1 has been released, bringing significant improvements to AMD’s RADV, Intel’s ANV, and NVIDIA’s NVK Vulkan drivers, including new API extensions and video encoding features, with a stable release expected in February 2026. <a href="https://app.daily.dev/posts/xa7zoapmT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare fixes WAF bypass vulnerability</strong>: Cloudflare patched a vulnerability in its Web Application Firewall (WAF) ACME validation logic that allowed attackers to bypass security rules and access origin servers directly by exploiting improper validation of ACME challenge tokens. <a href="https://app.daily.dev/posts/PoZiSIcPJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Fortinet FortiGate devices targeted via SSO exploitation</strong>: Arctic Wolf observed an automated cyberattack campaign exploiting SSO accounts on Fortinet FortiGate devices, leading to unauthorized configuration changes, persistence account creation, VPN access, and data exfiltration, resembling a December 2025 campaign targeting SSO authentication bypass vulnerabilities. <a href="https://app.daily.dev/posts/R9RchEmNx" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>jQuery 4.0 drops legacy browser support, adds security features</strong>: jQuery 4.0, released after nearly a decade, removes support for Internet Explorer 10 and Edge Legacy, integrates Trusted Types and Content Security Policy, and migrates to ES modules, resulting in a smaller, more modern library. <a href="https://app.daily.dev/posts/uMYmsakGM" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Palantir CEO claims AI will eliminate need for immigration</strong>: Alex Karp, Palantir’s CEO, at the World Economic Forum, declared that AI will make large-scale immigration unnecessary in Western economies, except for specialized skills. This bold prediction comes despite industry giants like Gartner and PwC reporting that most companies see zero payoff from their AI investments, signaling a ‘trough of disillusionment.’ So, while Karp’s company supposedly saves clients 80% on costs, the rest of us are still waiting for AI to even pay for itself, let alone solve global labor markets. <a href="https://app.daily.dev/posts/3s1vrBGbY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI world models and game engines: a clash of the titans, or just more hype?</strong>: Apparently, we’re facing a grand design tension between traditional game engines, with their quaint notions of ‘performance’ and ‘efficiency,’ and ‘AI world models’ that prioritize ‘content creation speed’ over, you know, actual quality or determinism. The article hints at ‘learned hallucinations’ as a feature, not a bug, suggesting future hybrid systems will magically combine the best of both worlds. Because nothing says ‘innovation’ like a system that’s both inefficient and prone to making things up. <a href="https://app.daily.dev/posts/OIj9HQhHA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatGPT UI replication: looks good, breaks fast</strong>: ChatGPT can apparently mimic Apple’s UI demos visually, which is great for screenshots, but it completely misses the underlying system constraints. The generated implementations ‘break under performance load’ because, shocker, it doesn’t understand how SwiftUI actually manages UI. The solution? More ‘proper mentoring’ for ChatGPT. Because clearly, the AI just needs a stern talking-to about performance best practices, not a fundamental understanding of how software actually works. <a href="https://app.daily.dev/posts/LQOYGhWvV" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
