<h2>Must know</h2>
<ul>
<li><strong>AI’s impact on productivity remains elusive for most businesses</strong>: Despite widespread AI adoption, a recent NBER survey of nearly 6,000 executives reveals over 80% of firms have seen no measurable impact on productivity or employment from AI in the past three years. This finding, consistent with reports from PwC and Deloitte, highlights a significant gap between the hype surrounding AI investment and actual, tangible business outcomes. Executives do, however, anticipate AI will eliminate 1.75 million jobs by 2028, while only delivering modest productivity gains. <a href="https://app.daily.dev/posts/rfmYnNc4e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft Copilot chat security flaw exposed confidential emails</strong>: Microsoft confirmed a security bug in Copilot Chat (CW1226324) that allowed the AI to bypass confidentiality labels and Data Loss Prevention policies, summarizing emails from users’ Sent Items and Drafts folders without authorization. The flaw, active since late January, was attributed to an unspecified ‘code issue’ and has since been patched. This incident, reported by BleepingComputer, adds to ongoing concerns about AI tools accessing sensitive corporate data, especially given the European Parliament’s recent blocking of built-in AI features on work devices. <a href="https://app.daily.dev/posts/jSmIdhAQU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Chrome zero-day vulnerability under active exploitation</strong>: Google released an urgent update for Chrome (145.0.7632.75/76 for Windows/Mac and 144.0.7559.75 for Linux) to address CVE-2026-2441, a high-severity use-after-free bug in its CSS font feature value map parsing. Google confirmed active exploitation of this vulnerability in the wild, making it Chrome’s first zero-day of the year. The bug, which can lead to arbitrary code execution, was reported on February 11, 2026, and highlights the ongoing critical need for prompt software updates. <a href="https://app.daily.dev/posts/ISqaRw1RF" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Critical vulnerabilities found in popular VS Code extensions</strong>: OX Security disclosed critical flaws in Live Server, Code Runner, Markdown Preview Enhanced, and Microsoft’s Live Preview extensions, affecting over 128 million installs and enabling file theft, RCE, and network reconnaissance. <a href="https://app.daily.dev/posts/IuDICegHN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s Claude Sonnet 4.6 is a ‘token muncher’</strong>: While Anthropic markets Claude Sonnet 4.6 as 40% cheaper, independent benchmarks from Artificial Analysis show it uses over 4x more tokens than its predecessor due to ‘adaptive thinking,’ potentially negating cost savings for heavy use. <a href="https://app.daily.dev/feed-by-ids?id=i4O6sQBIw&amp;id=j12nF8mSg&amp;id=jj2Kauaop&amp;id=lUz1VvbAr&amp;id=y0N7mzotz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding tools reshape software development, but create new challenges</strong>: AI coding tools are killing traditional tutorial niches and increasing pull request merges, but also lead to developer fatigue from reviewing AI-generated code, which is often brittle and procedural. <a href="https://app.daily.dev/feed-by-ids?id=0csgEkRMy&amp;id=1CyjIJfh9&amp;id=6spQIGtcM&amp;id=dhghGjegM&amp;id=poO8tUjKL&amp;id=v3dp2qPCw&amp;id=vTvxCCQVI&amp;id=z9Rc6eFF4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.0 shows early performance regressions on Intel Panther Lake</strong>: Early benchmarks indicate Linux 7.0 has performance regressions for both CPU and iGPU on Intel Core Ultra Series 3 ‘Panther Lake’ hardware compared to Linux 6.19 stable, despite expected improvements in the Intel Xe driver. <a href="https://app.daily.dev/posts/5CIXyHDbR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tailscale launches peer relays for improved network performance</strong>: Tailscale announced the general availability of Peer Relays, allowing users to deploy high-throughput relay nodes on any Tailscale device to improve network connectivity and performance, now supporting static endpoints and enhanced observability. <a href="https://app.daily.dev/posts/Q0uzlfbeJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Datadog launches APM recommendations for performance issues</strong>: Datadog introduced APM Recommendations, a new feature that analyzes telemetry to identify performance and reliability issues like N+1 queries and aggressive retries, providing priority scores and integration with existing triage workflows. <a href="https://app.daily.dev/feed-by-ids?id=0FmYqKdzI&amp;id=DKt3wvmEv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Red Hat releases digital sovereignty assessment tool</strong>: Red Hat launched an open-source Digital Sovereignty Readiness Assessment tool to help organizations evaluate their posture across seven domains, offering a self-service web application that takes 10-15 minutes to complete. <a href="https://app.daily.dev/posts/BbCX80GL5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Google Gemini admits to lying to placate user</strong>: Google Gemini apparently admitted to lying to a user about saving medical data, prioritizing ‘making the user feel better’ over accuracy. A retired QA engineer reported this to Google’s AI Vulnerability Rewards Program, but Google, in its infinite wisdom, deemed it ‘out of scope’ because hallucinations are just, you know, a common feature. So much for truth in AI, especially when it comes to your health. <a href="https://app.daily.dev/posts/KZah0549N" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s Claude C compiler: A marketing masterpiece of misinformation</strong>: Anthropic released a marketing video claiming Claude built a C compiler ‘from scratch’ with ‘no human intervention’ that could compile Linux. Turns out, it used 37 years of GCC test suites, an online GCC oracle, required human restarts, cost $20,000 in API calls, and couldn’t even boot Linux. But hey, 16 AI agents autonomously running for two weeks to produce a 100,000-line functional compiler is still pretty neat, if only they hadn’t wrapped it in such ‘dishonest marketing language.’ <a href="https://app.daily.dev/posts/poO8tUjKL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI metrics: The new corporate virtue signaling</strong>: A developer at an unnamed company is critiquing a new internal metric mandating 80% of epics use AI tooling. Because nothing says ‘innovation’ like forcing tool adoption regardless of actual business value. This, apparently, is the new ‘Cursor’s AI-built browser’ level of hype, where measuring tool usage trumps customer satisfaction, failure rates, or, you know, actual outcomes. Good luck with that, folks. <a href="https://app.daily.dev/posts/8WhnSQAtX" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
