<h2>Must know</h2>
<ul>
<li><strong>Anthropic and OpenAI clash over AI ads</strong>: Anthropic is planning Super Bowl ads to mock OpenAI’s move to introduce advertising in ChatGPT, positioning its Claude AI as ad-free. OpenAI CEO Sam Altman responded on X, calling Anthropic’s ads dishonest and defending OpenAI’s advertising principles. This public spat highlights differing business models and competitive tension, with Anthropic relying on enterprise contracts while OpenAI reportedly burns billions annually. <a href="https://app.daily.dev/posts/TmM2mNSSP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI launches enterprise AI agent platform frontier</strong>: OpenAI has launched ‘Frontier,’ a platform for enterprises to build, deploy, and manage AI agents, treating them like employees with onboarding and feedback loops. Early customers include HP and Uber, aiming to connect siloed data and automate workflows across tools like Salesforce. This positions OpenAI to compete in the enterprise AI agent management space, shifting focus from base models to higher-value applications. <a href="https://app.daily.dev/posts/TGZyay3jK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent security vulnerabilities identified</strong>: Snyk’s research, ‘ToxicSkills,’ found critical security vulnerabilities in the AI Agent Skills ecosystem, with 13.4% of scanned skills containing critical issues and 76 confirmed malicious payloads. These vulnerabilities pose a supply chain risk for platforms like OpenClaw and Claude Code. Operant AI launched Agent Protector to provide real-time security for autonomous AI agents, offering shadow agent discovery and inline protection against privilege escalation and data exfiltration. <a href="https://app.daily.dev/feed-by-ids?id=0BdgmYNfg&amp;id=0e28QTlYb&amp;id=2vNCw5dB4&amp;id=dL9ys7HAk&amp;id=the2EenMO" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Anthropic releases Claude Opus 4.6 with agent teams</strong>: Anthropic launched Claude Opus 4.6, featuring ‘agent teams’ for parallel task execution, an expanded 1 million token context window, and new PowerPoint integrations, aiming to broaden its appeal beyond developers to knowledge workers. <a href="https://app.daily.dev/feed-by-ids?id=G7Sre82A1&amp;id=Ijw4mCG7B" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Laravel launches AI SDK for application development</strong>: Laravel released an official AI SDK in beta, providing a unified API for integrating AI capabilities from multiple providers like Anthropic, Gemini, and OpenAI into Laravel applications, supporting features like agents, memory, and structured outputs. <a href="https://app.daily.dev/posts/ND9RAWZ9f" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft appoints engineering quality czar</strong>: Microsoft CEO Satya Nadella appointed Charlie Bell as engineering quality czar to improve product quality across the board, addressing issues like Azure outages and low Copilot adoption, with Hayete Gallot returning to lead security. <a href="https://app.daily.dev/posts/NGytU6zYE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS launches new EC2 instances</strong>: AWS introduced new EC2 High Memory U7i-6TB instances with Intel Xeon Sapphire Rapids processors and new C8id, M8id, and R8id instances powered by custom Intel Xeon 6 processors, offering significant performance and memory bandwidth improvements. <a href="https://app.daily.dev/posts/JPt4mjvME" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Substack user data breach disclosed</strong>: Substack disclosed a data breach from October 2025 that exposed user email addresses, phone numbers, and internal metadata, remaining undetected for months, though passwords and financial information were not compromised. <a href="https://app.daily.dev/posts/myrJIoXYs" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI bot traffic surges, threatening publishers</strong>: AI bots now account for a significant portion of web traffic, primarily for retrieval augmented generation (RAG), leading to a sharp decline in clickthrough rates to publishers and an ‘arms race’ as publishers deploy defenses. <a href="https://app.daily.dev/posts/5O2tzW5oh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft develops LLM backdoor scanner</strong>: Microsoft developed a scanner to detect malicious backdoors in open-weight AI models by identifying attention hijacking patterns, data leakage, and fuzzy trigger matching, addressing a critical supply chain risk for enterprises using third-party LLMs. <a href="https://app.daily.dev/posts/O9jvh85A8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Elixir is the ‘best language for AI’ says unverified study</strong>: An article boldly claims Elixir is the ‘best language for AI,’ citing an unnamed ‘Tencent study’ that supposedly found Elixir achieved a 97.5% LLM completion rate. Apparently, Elixir’s immutability and ‘decade-long stability’ are the secret sauce, making it perfect for AI, because who needs actual evidence when you have a mysterious study and a strong pro-Elixir bias? <a href="https://app.daily.dev/posts/1D7tGyRwk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>UK deepfake detection plan criticized as insufficient</strong>: The UK government, in its infinite wisdom, announced plans to develop a framework for evaluating deepfake detection technologies, partnering with Microsoft and academics. Cybersecurity expert Dr. Ilia Kolochenko, however, quickly pointed out that this is about as useful as a chocolate teapot, arguing that the real problem isn’t detection, but enforcement. Because, you know, bad actors are just going to line up to have their deepfakes detected and removed. <a href="https://app.daily.dev/posts/EQCYBDiDB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are going rogue, survey says</strong>: A survey of IT executives claims 1.5 million AI agents are ‘at risk of going rogue’ because they lack monitoring or security oversight. Security experts are comparing this to the Titanic disaster, because apparently, an unmonitored script is now equivalent to an iceberg. It’s almost as if deploying ‘invisible’ AI agents via low-code tools without proper scrutiny might lead to, gasp, security issues. Who could have predicted such a thing? <a href="https://app.daily.dev/posts/RMzBzQnM3" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
