<h2>Must know</h2>
<ul>
<li><strong>OpenAI introduces ads and a new Go tier</strong>: OpenAI is rolling out targeted ads in the free and Go tiers of ChatGPT for US users, citing financial losses and the need to sustain free access, as reported by TechCrunch. Ads will appear at the bottom of conversations and are labeled, with OpenAI assuring they won’t influence chatbot answers or user data. The new Go tier, priced at $8/month, offers more features than the free version, positioning ads as a way to subsidize global accessibility. This marks a significant shift from CEO Sam Altman’s previous stance that ads would be a “last resort” for monetization. <a href="https://app.daily.dev/posts/DQh9IAXNl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Mandiant releases Net-NTLMv1 rainbow tables</strong>: Mandiant has publicly released rainbow tables capable of cracking Net-NTLMv1 authentication hashes in under 12 hours using consumer-grade hardware. This move aims to highlight the severe, long-known vulnerabilities of the deprecated Microsoft protocol and push organizations to finally disable it. Mandiant emphasizes that credential theft via Net-NTLMv1 is “trivial” and urges immediate action through Group Policy to mitigate risks, as detailed in their security advisory. <a href="https://app.daily.dev/posts/2q2phjvWS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI firms accelerate healthcare expansion amid concerns</strong>: Major AI players like OpenAI and Anthropic are rapidly expanding into healthcare, with OpenAI acquiring health startup Torch and Anthropic launching a specialized Claude AI model. This “AI healthcare gold rush” is attracting significant investment, but experts warn of critical risks including AI model hallucination, medical misinformation, and security vulnerabilities with sensitive patient data, according to industry reports. The rapid deployment raises questions about regulatory oversight and clinical efficacy. <a href="https://app.daily.dev/posts/YNV6XIM0c" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Nvidia DGX Spark brings powerful AI to the desktop</strong>: Nvidia’s DGX Spark is a new compact AI workstation, delivering one petaFLOP of performance with a GB10 Grace Blackwell Superchip, targeting AI developers for local prototyping and fine-tuning of large models without cloud dependencies. <a href="https://app.daily.dev/posts/Xk9eNeVDY" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Claude Code improves tool loading efficiency</strong>: Claude Code has implemented dynamic tool loading for MCP servers, reducing token usage by up to 95% and significantly improving tool selection accuracy by only loading relevant tools on demand, rather than preloading all definitions. <a href="https://app.daily.dev/feed-by-ids?id=WwZicivI3&amp;id=uq0isMFwF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google previews AI SQL generation in BigQuery</strong>: Google is testing a new “Comments to SQL” feature in BigQuery, using Gemini AI to convert natural-language instructions within SQL comments into executable queries, aiming to accelerate data analysis. <a href="https://app.daily.dev/posts/5AcBUpc2i" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LLM Compressor 0.9.0 enhances model quantization</strong>: LLM Compressor 0.9.0 introduces refactored attention and KV cache quantization with support for multiple schemes (FP4, INT8, FP8), a new ‘model_free_ptq’ path, and batched calibration for faster runtimes. <a href="https://app.daily.dev/feed-by-ids?id=Y9B2mh3yd&amp;id=YCoHMIgPO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloudflare boosts resilience and acquires Astro</strong>: Cloudflare launched its “Code Orange: Fail Small” initiative to prevent outages through phased rollouts and improved failure handling, while also acquiring Astro Technology Company to bolster its developer ecosystem. <a href="https://app.daily.dev/feed-by-ids?id=OdNHxfcwg&amp;id=yRAOJZmiR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open-source LLM observability gets a boost with Langfuse acquisition</strong>: ClickHouse acquired Langfuse, a leading open-source platform for LLM observability and prompt management, which will remain open-source and enhance ClickHouse’s Agentic Data Stack with AI quality monitoring. <a href="https://app.daily.dev/posts/7DDF3YQda" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go 1.26 RC2 released with security updates</strong>: Go 1.26 Release Candidate 2 is out, bringing unspecified security fixes for the archive/zip library and Request.ParseForm, with similar updates also released for Go 1.25.6 and Go 1.24.12. <a href="https://app.daily.dev/posts/DFYbtJ1rQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-generated browser: 3 million lines of code, 0 working product</strong>: Cursor’s CEO proudly announced that hundreds of GPT-5.2 agents generated over 3 million lines of code for a web browser in a week, claiming “meaningful progress.” Yet, the actual codebase doesn’t compile, has 34 errors and 94 warnings, and lacks any verifiable demo. It seems “AI-generated code” is still mostly “AI-generated marketing hype,” as one article points out, proving that quantity doesn’t equal quality, especially when it comes to production-ready software. <a href="https://app.daily.dev/feed-by-ids?id=UbZlY7YDE&amp;id=qHCb0ov8q&amp;id=7Sqt9u8B2&amp;id=V3KHuMgUU&amp;id=V42saD54U" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI and Microsoft sued by Elon Musk, again</strong>: A federal judge has decided that Elon Musk’s lawsuit against OpenAI and Microsoft, alleging they broke nonprofit commitments by going for-profit, will proceed to a jury trial. Because nothing says “mission-driven” like a multi-billion dollar legal battle over who gets to profit from AI. Apparently, the court will now decide if accepting billions from Microsoft constitutes a breach of “founding commitments.” Good luck with that, folks. <a href="https://app.daily.dev/posts/pwpD2QRjm" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Junior developers: Now an endangered species, thanks to AI</strong>: According to some, junior developer roles have plummeted by 46-67% because AI can now handle all the entry-level coding. So, instead of learning to code, aspiring developers should become “AI architects” who orchestrate AI systems and manage context windows. Because apparently, the future of software engineering is less about writing actual code and more about whispering sweet nothings into an LLM’s ear, hoping it spits out something coherent. What could possibly go wrong when the entire career ladder disappears? <a href="https://app.daily.dev/posts/dc9HNh5yX" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
