<h2>Must know</h2>
<ul>
<li><strong>AI model war heats up with new releases from Anthropic and OpenAI</strong>: Anthropic and OpenAI are in a full-blown AI model war, dropping new versions within an hour of each other. Anthropic’s Claude Opus 4.6, available in Microsoft Foundry, boasts a 1M token context window and excels in complex coding and agentic workflows. OpenAI’s GPT-5.3 Codex is a first-party coding agent app, bringing them up to speed with competitors. One developer even used 16 Claude Opus 4.6 instances to autonomously build a 100,000-line Rust-based C compiler in two weeks for $20,000, achieving a 99% pass rate on GCC torture tests, according to Anthropic researcher Nicholas Carlini. This rapid release cycle and focus on agentic capabilities signal a major shift in how AI is being used for development. <a href="https://app.daily.dev/feed-by-ids?id=6vLfIe4vv&amp;id=8HMA0hoKL&amp;id=B4HtIccoU&amp;id=BQ8poB6iB&amp;id=CWUtlxYzI&amp;id=D8NcAVvwq&amp;id=EmRFinQYK&amp;id=LVl8sC9Gk&amp;id=YIoiLTXiE&amp;id=gP03wq1GE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft unveils Rust-based sandboxing project LiteBox</strong>: Microsoft open-sourced LiteBox, an experimental library OS written in Rust, designed for application sandboxing with a minimal attack surface. Unlike containers or VMs, it uses a library OS model for stronger isolation without heavy overhead. It supports AMD SEV-SNP for confidential computing and can run unmodified Linux applications. This is a significant move for security, especially for those looking for robust isolation in their applications, though Microsoft notes it’s not production-ready yet. <a href="https://app.daily.dev/feed-by-ids?id=jIlb9nmzq&amp;id=0PdnWmIzP&amp;id=3E2UGxSLK&amp;id=7XjDZBX9G&amp;id=8NUCb75wT&amp;id=8U9kW84ss&amp;id=CtbDgV4IB&amp;id=CyiEJWnVI&amp;id=HbVsLg8eZ&amp;id=HjXt5zm3t" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI accelerates AWS environment breach to 8 minutes</strong>: Threat actors breached an AWS environment, escalating from initial access to full admin privileges in just 8 minutes by leveraging large language models (LLMs). The attack started with credentials found in misconfigured public S3 buckets and involved Lambda function code injection for privilege escalation. This incident, reported by AWS, highlights how AI can accelerate cyberattacks, automating reconnaissance, code generation, and real-time decision-making. It’s a stark reminder that misconfigurations, even seemingly minor ones, can be exploited with alarming speed when AI is in the mix. <a href="https://app.daily.dev/posts/6mJxFRcx6" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>State of JavaScript 2025 survey results released</strong>: The State of JavaScript 2025 survey results are out, offering insights into the latest trends, tools, and developments in the JavaScript ecosystem. <a href="https://app.daily.dev/posts/lJAPeqJsk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Snowflake integrates pg_lake for open data standards</strong>: Snowflake integrated <code>pg_lake</code>, an open-source PostgreSQL extension, into its Snowflake Postgres service, allowing direct work with data lakehouses and Apache Iceberg tables via SQL. <a href="https://app.daily.dev/posts/kBWxo6cQz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Interop 2025 achieves 97% pass rate for web standards</strong>: Interop 2025 concluded with a 97% pass rate across all browsers, showing significant progress in web platform interoperability across 19 focus areas like CSS Anchor Positioning and View Transitions. <a href="https://app.daily.dev/posts/X7hlaiKma" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Qodo launches AI-powered code review platform 2.0</strong>: Qodo launched version 2.0 of its AI-powered code review platform, introducing multiple specialized AI agents that use recall and memory to review code at a senior engineer level. <a href="https://app.daily.dev/feed-by-ids?id=4Pp2U3N5Y&amp;id=PAc5sU2Jg" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ubuntu expands RISC-V support with K3 and K1 processors</strong>: Canonical will support SpacemiT K3 RISC-V processor with Ubuntu 26.04 LTS, targeting one of the first commercially available RVA23-compliant chips with security extensions and AI-focused vector computing. <a href="https://app.daily.dev/posts/C9IUbBaRM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PureRef 2.1 update brings new features and languages</strong>: PureRef released version 2.1, its first update since 2024, adding a background grid with snap-to-grid, shape drawing tools, and support for six new languages. <a href="https://app.daily.dev/feed-by-ids?id=EOJeYNj9E&amp;id=L091whE3h&amp;id=M4E3Sbe0D&amp;id=lA6Q9OrWc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go and Python ecosystems see updates and integrations</strong>: Microsoft released new builds of Go versions 1.25.7-1 and 1.24.13-1, including security fixes, while the Python ecosystem explores ML alternatives and Rust integrates with Python via PyO3. <a href="https://app.daily.dev/feed-by-ids?id=27vA6E5rB&amp;id=3p16HuQId&amp;id=42OXMTBFy&amp;id=6lrbyyyQ3&amp;id=8gkBetvnO&amp;id=WFUKfQ0NK&amp;id=cAgN2AojU" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Pinterest fires engineers for checking layoff list</strong>: Pinterest, in its infinite wisdom, laid off 15% of its workforce due to ‘AI restructuring’ but refused to disclose who was terminated. When two engineers dared to write a script to figure it out, they were promptly fired for accessing ‘confidential information.’ Because apparently, knowing if your colleagues still have jobs is a state secret. The CEO then had the audacity to tell remaining employees to align or leave. Nothing says ‘we value our people’ like firing them for wanting basic transparency. <a href="https://app.daily.dev/posts/Om0rqgwIK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code floods open source, breaks hosting bills</strong>: Apparently, AI-generated code is now so prolific and low-quality that open-source projects are being forced to close external pull requests. One developer’s hosting bill skyrocketed from $30 to $2,000 because Meta’s crawler hit their Vercel site 11 million times in a month. So, while AI makes code generation ‘cheap,’ it seems the cost of dealing with its garbage and the operational overhead of serverless platforms is just shifting to maintainers and developers. Who needs quality when you have quantity, right? <a href="https://app.daily.dev/posts/09F3bx5zo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>FBI can’t crack iPhone with Lockdown Mode enabled</strong>: The FBI’s elite Computer Analysis Response Team (CART) was apparently stumped by an iPhone with Apple’s Lockdown Mode enabled, unable to extract data during a raid on a Washington Post reporter. This, of course, proves the mode’s effectiveness against law enforcement. So, while everyone else is worried about AI taking over the world, at least we know our iPhones can still keep a secret from the feds. Good to know some tech actually works as advertised for privacy, even if it’s just to annoy the authorities. <a href="https://app.daily.dev/posts/PXLqFkE9O" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
