<h2>Must know</h2>
<ul>
<li><strong>AI agent security vulnerabilities exposed</strong>: Snyk’s ‘ToxicSkills’ research found critical security flaws in 13.4% of AI agent skills, with 76 confirmed malicious payloads designed for credential theft and data exfiltration. This highlights a significant supply chain risk for enterprises using third-party LLMs, especially with platforms like OpenClaw and Claude Code. Operant AI launched Agent Protector to offer real-time security for autonomous AI agents, focusing on 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>
<li><strong>Anthropic and OpenAI clash over AI ads</strong>: Anthropic plans Super Bowl ads mocking OpenAI’s move to introduce advertising in ChatGPT, positioning its Claude AI as ad-free. OpenAI CEO Sam Altman called Anthropic’s ads ‘dishonest’ and ‘authoritarian,’ defending OpenAI’s transparent ad implementation. This public spat highlights differing business models and intense rivalry, with Anthropic relying on enterprise contracts while OpenAI reportedly burns $9 billion annually. <a href="https://app.daily.dev/posts/TmM6mNSSP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Global cyberespionage campaign targets critical infrastructure</strong>: A sophisticated nation-state group, TGR-STA-1030, has been conducting a global cyberespionage campaign since January 2024, compromising government and critical infrastructure in 37 countries. The group uses phishing, N-day exploits, and a new Linux kernel rootkit, ShadowGuard, to exfiltrate sensitive data. Unit 42 researchers uncovered the operations, and CISA is aware of the threat. <a href="https://app.daily.dev/posts/KNu0fQxJH" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI coding agents advance software development</strong>: AI coding agents are automating tasks from feature requests to deployment, with Anthropic’s Claude Code introducing ‘agent teams’ for parallel work and Cursor’s multi-agent system achieving ~1,000 commits per hour. <a href="https://app.daily.dev/feed-by-ids?id=I2lSMdhWD&amp;id=Q3sR9E9tC&amp;id=WcVKYHvlb&amp;id=bgc1cUCVA&amp;id=bq4ihvY3K&amp;id=dsKDEbeOE&amp;id=RtOeT8YDy&amp;id=YPOCeTWKC&amp;id=gdolSqfBx&amp;id=kPp1hKs76" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic releases Claude Opus 4.6 with expanded context</strong>: Anthropic launched Claude Opus 4.6, now available on Amazon Bedrock, Vercel’s AI Gateway, and Google’s Vertex AI, featuring a 1 million token context window, ‘agent teams,’ and enhanced Excel/PowerPoint integrations. <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>Microsoft launches enterprise AI agent platform ‘Frontier’</strong>: OpenAI launched ‘Frontier,’ an end-to-end platform for enterprises to build, deploy, and manage AI agents, allowing programming of agents to connect to external data and applications with access controls. <a href="https://app.daily.dev/posts/TGZyay3jK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Laravel releases official AI SDK</strong>: Laravel launched an official AI SDK in beta, providing a unified API for integrating AI capabilities into Laravel applications, supporting multiple AI providers like Anthropic, Gemini, OpenAI, and ElevenLabs. <a href="https://app.daily.dev/posts/ND9RAWZ9f" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AWS launches new EC2 instances</strong>: AWS launched EC2 High Memory U7i-6TB instances with custom Intel Xeon Scalable 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>ScyllaDB integrates USearch for vector search</strong>: ScyllaDB now includes vector search capabilities using the open-source USearch library, allowing storage of vector embeddings alongside structured data and leveraging ScyllaDB’s shard-per-core architecture for optimized performance. <a href="https://app.daily.dev/posts/pKNE7iWlw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GNOME Shell and Mutter 50 beta released</strong>: GNOME Shell and Mutter 50 beta releases introduce stable Variable Refresh Rate (VRR) support, improved frame scheduling, and enhanced discrete GPU detection, paving the way for the stable GNOME 50 release in March. <a href="https://app.daily.dev/posts/buHOzLdcP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI bot traffic surges, threatening publishers</strong>: AI bots now account for one visit for every 31 human visits, primarily driven by RAG bots, causing clickthrough rates from AI applications to source websites to plummet to 0.27%. Publishers are deploying ‘increasingly aggressive defenses’ against these bots, which are ‘developing sophisticated tactics to bypass website protections.’ So, basically, the internet is becoming a battleground between AI and the content creators it’s supposedly ‘helping.’ <a href="https://app.daily.dev/posts/5O2tzW5oh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Unmonitored AI agents pose significant security risk</strong>: A survey of IT executives claims 1.5 million AI agents lack active monitoring, leading to 88% of executives reporting AI agent-related security incidents. Security experts are comparing this to the ‘Titanic disaster,’ warning that ‘invisible’ AI agents deployed via low-code tools are bypassing traditional IT scrutiny. Because, apparently, we needed another thing to keep CISOs up at night, and ‘rogue’ AI agents sound much more dramatic than ‘poorly configured automation scripts.’ <a href="https://app.daily.dev/posts/RMzBzQnM3" target="_blank" rel="noopener nofollow">Read more</a></li>
<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. The article touts Elixir’s immutability, stability, and low operational complexity as key benefits, conveniently overlooking any potential drawbacks or the fact that this ‘study’ seems to exist only in the article itself. Sounds like someone’s trying to make their favorite language the next big thing in AI, regardless of actual evidence. <a href="https://app.daily.dev/posts/1D7tGyRwk" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
