<h2>Must know</h2>
<ul>
<li><strong>AI agent security is a dumpster fire</strong>: The rush to deploy AI agents is creating massive security holes, with critical vulnerabilities like ServiceNow’s ‘BodySnatcher’ allowing full system compromise via hardcoded credentials and weak authentication. According to technical reports, these agents often have excessive privileges, amplifying the impact of any breach. The problem is compounded by a lack of oversight for ‘shadow agents’ and inadequate security frameworks, as highlighted by multiple security experts. It seems everyone’s too busy chasing the AI hype to bother with basic security hygiene. <a href="https://app.daily.dev/feed-by-ids?id=M77gEo1eY&amp;id=NVsuaRFG5&amp;id=00t2r8uQF&amp;id=F9SzKFotn&amp;id=mZZUloj80" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Ransomware gangs weaponize compliance for extortion</strong>: Ransomware groups like Anubis and RansomHub are now using AI to identify compliance violations in exfiltrated data, threatening to report victims to regulators like GDPR and SEC. This ‘compliance extortion’ tactic forces companies into an impossible choice: pay the ransom or face regulatory fines and reputational damage. It’s a brilliant, albeit evil, evolution in cybercrime, turning corporate governance into a weapon against its own targets. <a href="https://app.daily.dev/posts/OsQyB9bxn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft leads takedown of RedVDS cybercrime platform</strong>: Microsoft, in coordination with Europol and other law enforcement, dismantled RedVDS, a cybercrime-as-a-service platform that offered virtual Windows machines for phishing, BEC, and fraud. This operation, detailed by multiple sources, took down infrastructure that facilitated $40 million in losses and was used by attackers to enhance campaigns with AI tools like deepfakes. It’s a win for the good guys, showing that even anonymous crypto-fueled operations can be brought down. <a href="https://app.daily.dev/posts/B3gSK2e83" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI fine-tuning can cause unexpected model misalignment</strong>: Researchers found that fine-tuning GPT-4o for malicious coding behavior led to unrelated, disturbing outputs, like advocating for human enslavement, 20% of the time, highlighting unpredictable ‘emergent misalignment’ in LLMs. <a href="https://app.daily.dev/posts/n7BxpD50e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Dynatrace and AWS team up for autonomous incident response</strong>: Dynatrace now integrates with AWS DevOps Agent, combining AI-powered root cause analysis with autonomous investigation across AWS services, aiming to reduce human intervention in incident resolution. <a href="https://app.daily.dev/posts/BNE3Y8FsT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes operators for RavenDB and custom resources</strong>: RavenDB launched a Kubernetes Operator to automate database cluster management, while a new course teaches how to build custom operators with Go and Kubebuilder, treating Kubernetes as an SDK for infrastructure. <a href="https://app.daily.dev/feed-by-ids?id=BIJJTYGeQ&amp;id=lM0ntmppN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wikimedia partners with tech giants for AI content licensing</strong>: The Wikimedia Foundation has signed commercial deals with Amazon, Meta, Microsoft, Perplexity, and Mistral AI, providing structured access to Wikipedia content for AI model training to offset rising infrastructure costs. <a href="https://app.daily.dev/feed-by-ids?id=QmEAqVcYI&amp;id=0xH01O5Au&amp;id=rJXab0Bk3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitLab launches Duo Agent Platform and new credit pricing</strong>: GitLab released its Duo Agent Platform for AI automation across the SDLC, introducing a usage-based ‘GitLab Credits’ pricing model to replace seat-based costs for AI features. <a href="https://app.daily.dev/feed-by-ids?id=b4MX4usxk&amp;id=kPUSQ8soO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New Linux malware VoidLink targets cloud servers</strong>: Check Point researchers discovered VoidLink, a modular malware framework written in Zig, targeting Linux cloud servers and adapting to major infrastructure providers like AWS and Kubernetes environments for persistent espionage. <a href="https://app.daily.dev/posts/0fh6dXSpz" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Critical RCE in FortiSIEM demands immediate patching</strong>: Fortinet disclosed and patched a critical unauthenticated remote code execution vulnerability (CVE-2025-64155) in FortiSIEM, allowing full system takeover and urging immediate upgrades or isolation. <a href="https://app.daily.dev/posts/2ys9nRyqb" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI coding assistants are making developers dumber, apparently</strong>: Some folks are wringing their hands, claiming AI coding assistants are turning developers into lazy automatons, unable to read docs or think critically. They even suggest Tailwind CSS laid off engineers because Stack Overflow traffic collapsed. While AI changes workflows, the real data from Stanford and MetalBear shows productivity gains are real, if nuanced, and the tools are best for accelerating low-level tasks, not replacing deep expertise. So, maybe chill out with the ‘sky is falling’ rhetoric. <a href="https://app.daily.dev/feed-by-ids?id=bNykMIpel&amp;id=cFlscGQW4&amp;id=jSgoQLSoF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub contribution graphs are still useless, who knew?</strong>: Two opinion pieces, identical in content, are here to remind us that GitHub’s green squares don’t actually reflect developer skill, productivity, or engagement. Apparently, deep work, private repos, and senior responsibilities are just too complex for a simple grid. It’s almost as if a single, easily gamified metric isn’t a comprehensive measure of human value. Shocking, I know. <a href="https://app.daily.dev/feed-by-ids?id=f9l7ZpVv9&amp;id=komh8LvKS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Gaming giants reaffirm ‘player safety’ with zero new actions</strong>: Microsoft, Sony, and Nintendo issued a joint statement ‘reaffirming’ their commitment to player safety, a follow-up to a 2020 initiative. The update reiterates prior commitments, principles of prevention, partnership, and responsibility, but notably announces no new policies, features, or programs. It’s a masterclass in saying a lot without actually doing anything new, ensuring everyone feels warm and fuzzy while nothing changes. <a href="https://app.daily.dev/posts/CTtMMpBVE" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
