<h2>Must know</h2>
<ul>
<li><strong>Mandiant drops NTLMv1 rainbow tables, urges immediate deprecation</strong>: Mandiant has publicly released rainbow tables capable of cracking Net-NTLMv1 password hashes in under 12 hours using consumer hardware. This move, hosted on Google Cloud, aims to force organizations to finally deprecate the 20-year-old, vulnerable protocol, which Mandiant says is still widely used despite known risks. They emphasize that credential theft is trivial when NTLMv1 is in play and provide guidance for detection and disabling, as reported by Mandiant itself. <a href="https://app.daily.dev/posts/2q2phjvWS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Major vendors ditch MITRE ATT&amp;CK evaluations</strong>: Microsoft, SentinelOne, and Palo Alto Networks have pulled out of the 2025 MITRE ATT&amp;CK Evaluations, citing resource prioritization for product development. MITRE’s CTO admitted the 2025 tests are more demanding, now including cloud environments and alert volume tracking. Industry experts are calling the evaluations ‘vendor theater,’ suggesting they serve PR more than genuine security improvement, according to CSO Online. <a href="https://app.daily.dev/feed-by-ids?id=c0ZYfkcgP&amp;id=hu11lcW1J" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI and non-human identity management for cloud security</strong>: Managing Non-Human Identities (NHIs) is critical for cloud data security, requiring a comprehensive, AI-driven lifecycle approach from discovery to remediation. AI enhances threat detection through pattern analysis and continuous monitoring, reducing data breach risks and improving compliance. CISOs need to ensure alignment across development, security, and R&amp;D teams to close potential gaps, as highlighted by Software Testing Magazine. <a href="https://app.daily.dev/feed-by-ids?id=lq3eNhRBB&amp;id=rc0IgUSh5" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>NVIDIA expands AI gaming and streaming with DLSS 4.5 and G-SYNC Pulsar</strong>: NVIDIA announced DLSS 4.5 with Dynamic Multi Frame Generation for 240+ FPS path-traced gaming on RTX 50 Series GPUs, alongside G-SYNC Pulsar monitors offering over 1,000Hz effective motion clarity. <a href="https://app.daily.dev/posts/jXjzYY9vI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI product launches and strategic partnerships in virtual assistants</strong>: Claude released Co-work for desktop automation, Google introduced Personal Intelligence for Gemini, and Apple partnered with Google to use Gemini models in Siri, intensifying competition in the AI assistant ecosystem. <a href="https://app.daily.dev/posts/3e5iSak1o" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Black Basta ransomware leader added to most wanted list</strong>: Ukrainian and German authorities identified Oleg Nefedov as the alleged leader of the Black Basta ransomware group, adding him to the EU Most Wanted list and an INTERPOL Red Notice after leaked chat logs revealed his leadership. <a href="https://app.daily.dev/posts/O0PAVFWl9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GPU supply constraints and AI export tariffs impact market</strong>: Nvidia and Asus clarified that RTX 5070 Ti and 5060 Ti GPUs are not cancelled, but supply shortages are projected to continue until 2027, while US export controls and tariffs have significantly reduced Nvidia’s market share in China. <a href="https://app.daily.dev/feed-by-ids?id=1QuQlmOmw&amp;id=onISZaYqR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LLMOps course focuses on production application of LLMs</strong>: A new LLMOps course emphasizes strategies for configuring and orchestrating large language models in production, covering decoding, generation parameter tuning, lifecycle management, and evaluation system design. <a href="https://app.daily.dev/posts/1q51Gs3MM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA boosts local AI video generation on RTX GPUs</strong>: NVIDIA announced significant local performance upgrades for AI video/image generation on RTX GPUs, including 3x faster video generation and 60% VRAM reduction with NVFP4/FP8 quantization and Lightricks’ LTX-2 model for 4K video. <a href="https://app.daily.dev/posts/JItr1KihQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong><a href="http://LWN.net" target="_blank" rel="noopener nofollow">LWN.net</a> hit by DDoS attack, Snapdragon X2 Elite laptop tested</strong>: <a href="http://LWN.net" target="_blank" rel="noopener nofollow">LWN.net</a> is experiencing a severe DDoS attack from tens of thousands of scrapers, while a pre-production Asus Zenbook A16 with a Snapdragon X2 Elite processor was tested at CES, showing promise despite some bugs. <a href="https://app.daily.dev/feed-by-ids?id=hTdHReUkc&amp;id=hZ8xhtHq8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-generated code: the gift that keeps on giving… maintainers headaches</strong>: Multiple engineering teams are reporting that AI-generated code, while compiling, often lacks architectural context and produces poor state management, leading to major refactoring that can take longer than manual implementation. Apparently, trusting plausible-looking AI outputs without expert oversight is a bad idea, as ChatGPT has been caught generating ‘realistic’ but non-existent API elements, according to various engineering team reports. So much for AI replacing developers; it seems it’s just creating more work for them. <a href="https://app.daily.dev/feed-by-ids?id=STWE5tRXe&amp;id=UAphs9F5B&amp;id=euzGVQGzQ&amp;id=gMHZMigM0&amp;id=m6P4U0J0b&amp;id=wCQfiCXDa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents: old wine in new bottles, with a side of hype</strong>: Apparently, ‘AI agents’ are the hot new thing, but multiple sources are pointing out that these ‘novel technologies’ are just a rebranding of APIs, web automation, and semantic web standards that have been around for over a decade. It seems the real ‘advancements’ are focused on building machine-readable APIs, not just slapping the term ‘agentic technology’ on everything to attract new investment rounds. So, while developers are enthusiastically building Pokémon Trainer AIs and using Claude Code as a ‘collaborative partner,’ the cynical among us are just seeing dollar signs and recycled concepts. <a href="https://app.daily.dev/feed-by-ids?id=4rJ96gwiU&amp;id=6IAZSWIz1&amp;id=6XeVWOBXO&amp;id=77jDjhI5C&amp;id=sH4hATTfN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Specialized AI agents: the ‘digital factory’ of tomorrow, or just more buzzwords?</strong>: Research is now claiming that large language models are ‘economically unsustainable’ and the future is a ‘digital factory’ of small, cheap specialized language models. Bold claims of democratizing AI and reducing costs by ‘several orders of magnitude’ are being thrown around, conveniently without attribution to specific companies or empirical deployment cases. NVIDIA’s survey of retail and CPG sectors, however, reports 91% of companies are using or exploring AI and 89% have seen revenue increases, so maybe the ‘economic unsustainability’ is just for those who aren’t buying NVIDIA’s blueprints. <a href="https://app.daily.dev/feed-by-ids?id=J1BKrq2u6&amp;id=J1IZlkzNv&amp;id=Yuaq0Gyje&amp;id=dliP781v1" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
