<h2>Must know</h2>
<ul>
<li><strong>Google’s gemini AI faces extraction attempts and nation-state misuse</strong>: Google has blocked over 100,000 attempts to extract proprietary reasoning capabilities from its Gemini AI model using distillation techniques, which it considers intellectual property theft. Simultaneously, Google’s threat report reveals nation-state actors from China, Iran, North Korea, and Russia are integrating Gemini into attack operations, including social engineering and phishing, with new malware families like HONESTCUE embedding Gemini’s API to generate malicious code that bypasses safety filters. This highlights the dual challenge of protecting advanced AI and preventing its weaponization. <a href="https://app.daily.dev/posts/OQLlwLpSw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic raises $30 billion, AI investment bubble concerns grow</strong>: Anthropic closed a massive $30 billion Series G funding round, doubling its valuation to $380 billion, driven by enterprise adoption of its Claude AI models and developer tools. Despite this, some experts, like Stephanie Kirmer, Staff ML Engineer, view the $200 billion AI investment as an unsustainable bubble, advocating for more realistic promises from AI companies. This raises questions about the long-term sustainability of current AI valuations and the potential for a market correction. <a href="https://app.daily.dev/feed-by-ids?id=0dRrHpMfv&amp;id=2KG0fxG7Z&amp;id=KBiI9wOrx&amp;id=MPMZLgvp3&amp;id=PryOnuBPQ&amp;id=Qmgd9m7GA&amp;id=s5BDVI2HR&amp;id=vELqgGbwK&amp;id=vz4jhHvn8&amp;id=xNryU16Jc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Odido data breach affects 6.2 million customers</strong>: Dutch mobile operator Odido disclosed a data breach impacting 6.2 million customers, roughly one-third of the Netherlands’ population. Attackers accessed their customer contact system, stealing names, addresses, phone numbers, dates of birth, bank account numbers, and ID document details. While passwords and call records were not compromised, customers are warned about potential phishing and impersonation scams, underscoring the ongoing vulnerability of large-scale customer data. <a href="https://app.daily.dev/posts/3j2lBB9LH" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Minimax launches cost-effective M2.5 AI model</strong>: Minimax released its M2.5 frontier AI model, priced at approximately $1/hour, claiming it’s over 90% cheaper than competitors like Claude Opus or GPT-5, and benchmarked as a top open model for coding tasks. <a href="https://app.daily.dev/posts/VQt2ETdNG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI governance gaps emerge with autonomous agents</strong>: AI agents and copilots in ERP, finance, and SaaS systems create governance gaps as they operate autonomously with sensitive data, necessitating a unified control plane to manage AI as both an identity and data problem. <a href="https://app.daily.dev/feed-by-ids?id=2Nhsskkxs&amp;id=7qf6CcnTl&amp;id=C2t5MyKnp&amp;id=EvW1FrjUS&amp;id=FyP2Wb8Vm&amp;id=IQCKAyNnH&amp;id=jvGGx2ulB&amp;id=lux1oBTdQ&amp;id=mcpWOltJw&amp;id=pTLCg9UIw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenClaw autonomous AI agent framework gains rapid popularity</strong>: OpenClaw, an open-source autonomous AI agent framework, has rapidly gained nearly 200k GitHub stars, offering continuous operation, persistent memory, and integration with platforms like Discord and Telegram, while allowing users full data ownership. <a href="https://app.daily.dev/posts/Mx18Wv9Wt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI reshapes software engineering roles</strong>: AI tools are shifting software engineering from routine coding to problem-solving, architecture, and system design, making judgment and understanding tradeoffs critical skills, while junior developers face challenges as entry-level grunt work disappears. <a href="https://app.daily.dev/feed-by-ids?id=0wFgrbFqP&amp;id=DcSKM9BQA&amp;id=GITxcoUi5&amp;id=GManH4sq3&amp;id=HHMxTl4yN&amp;id=RUKLi7SlQ&amp;id=ZY9MBJa3v&amp;id=buZd5qJnM&amp;id=cnjrcboRc&amp;id=i82gqqFnS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Spatie releases Laravel-PDF v2 with Cloudflare API driver</strong>: Spatie launched Laravel-PDF v2, featuring a Cloudflare browser rendering API driver that enables PDF generation without local dependencies like Node.js or Docker, alongside new features like queued PDF generation and metadata support. <a href="https://app.daily.dev/posts/cK3euMd6K" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Auth0 updates B2B plans for flexibility</strong>: Auth0 announced updates to its B2B pricing plans, including one Enterprise Connection for Free plan POCs and incremental M2M token add-ons for the B2B Essentials plan, aiming to enhance flexibility and scalability. <a href="https://app.daily.dev/posts/UzL5CQ1eX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Digitizing childhood memories with custom AI pipeline</strong>: A developer built a custom pipeline to digitize 200+ hours of childhood Video8 footage, using FireWire capture, AI-assisted transcription (WhisperX), HLS conversion, and a family streaming platform with Cloudflare R2 and Supabase, costing only ~$5/month. <a href="https://app.daily.dev/posts/zquW7brd7" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Gartner predicts 40% of AI projects will fail by 2027</strong>: Gartner, in its infinite wisdom, predicts 40% of AI projects will fail by 2027 due to ‘unsustainable velocity’ and ‘fragmented systems.’ Apparently, ‘shadow AI’ is also creating security vulnerabilities. Who knew that throwing AI at every problem without a coherent strategy might not work out? Shocking, I tell you. <a href="https://app.daily.dev/posts/ZY9MBJa3v" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI’s chatgpt ads: the slippery slope to monetization</strong>: OpenAI is now putting ads in ChatGPT, starting with ‘benign’ suggestions below responses. But according to Matt Wolfe, the real fun begins post-IPO when revenue pressure mounts. Get ready for ads to become less distinguishable or, even better, embedded directly into your AI-generated content. Because nothing says ‘user experience’ like a sponsored hallucination. <a href="https://app.daily.dev/posts/NO2cmpN4i" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s $20k c compiler: marketing hype or actual progress?</strong>: Anthropic’s claim of an AI-built C compiler, created by 16 Claude Opus agents for $20,000, is being called out as marketing hype. Apparently, it failed basic tests without manual configuration and relied on GCC for assembly/linking. So, it’s less ‘AI-built’ and more ‘AI-assisted, with a lot of human hand-holding and existing tools.’ But hey, the headline was good, right? <a href="https://app.daily.dev/posts/mZHhw9XzP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
