WordPress Vulnerability, AI Power Plays, and Netflix Cuts Costs
WordPress has a critical API vulnerability, China and the US are vying for AI power, and Netflix shows how generative AI is changing production.
Inhaltsverzeichnis
Today is one of those days when the AI world is shifting on several levels at once: security flaws in the mass-market WordPress system, new power shifts among models, and ever-stronger geopolitical competition for AI infrastructure. Add to that concrete real-world examples from media production and the military — so quite a lot of material for a Monday that reads like a conference with too little coffee.
🔐 Critical WordPress Vulnerability Allows Code Injection via API
A critical vulnerability has appeared in WordPress that, under certain conditions, even allows attackers to inject code. Particularly serious: According to heise, the flaw arises from a combination of SQL injection and an API error. This is exactly the kind of chain reaction where a “small” bug can become a full compromise.
Why does this matter? WordPress is still the backbone of a large share of the web — from hobby blogs to corporate websites. When a vulnerability like this becomes public, the window between patch and attack is often short. According to the report, WordPress has already released an update; the researchers who found the issue are also offering a hotfix. If you run WordPress, you should not wait for “later today” now. In security, later is usually just another word for “too late.”
🧠 Kimi K3 Puts the Western AI Consensus to the Test
With Kimi K3, Moonshot AI from China is once again causing unrest in the global AI market. According to initial assessments, the model is said to be on the level of Anthropic’s Opus 4.8 — and with a team of only around 300 people. The fact that even OpenAI strategist Dean W. Ball describes the model as “very good” shows that this is not just marketing, but real competition.
The bigger story behind this is geopolitical. Kimi K3 reopens the old debate: in the end, does compute power matter most, or can very strong models also be built with more efficient approaches? And what does that mean for US export controls if Chinese labs apparently continue to advance quickly? For you, this means the lead in the AI market is less stable than some in Silicon Valley like to claim. Or put differently: the apple doesn’t only fall from the tree in California.
🎬 Netflix Uses AI in 300 Productions
Netflix is now using generative AI in around 300 productions, mainly in post-production, as The Decoder reports. Co-CEO Ted Sarandos gives a concrete example: the docuseries “The American Experiment” includes 17 minutes of AI-generated material and was completed in half the time at half the cost.
This matters for the industry because Netflix is not just experimenting — it is already using AI in a scalable production environment. Important here is also the context: according to Sarandos, the money saved will not be used to reduce budgets, but to create more content. For media companies and creative teams, this is a clear signal: generative AI is no longer just a tool for early drafts, but is increasingly becoming part of the production chain. If you work in the industry, you should probably ask which tasks can be sensibly automated — and which cannot.
⚓ US Military Chooses AI-First Over Perfection
The US Navy Department is now pursuing a strategy that you might kindly call “bold” and less kindly call “deploy first, pray later.” According to The Decoder, the plan involves the “weaponization” of data and AI, including an AI-first approach for the fleet. Large language models are supposed to run directly on warships, and an AI war council is prioritizing use cases.
The core argument is remarkable: the risks of acting too slowly are greater than the risks of imperfectly aligned systems. That is a sentence that usually causes pain in regulated industries — and rightly so. But for the AI debate, it makes very clear where the discourse is heading: no longer “whether” AI enters critical systems, but “how fast.” The result: higher speed, but also less patience for perfect safety. A classic from the engine room of modern technology policy.
🏛️ Berlin and Paris Want Their Own Digital Backbone
Germany and France want to further reduce their dependence on US technology and are planning their own digital backbone, as heise reports. At the center are new AI security institutes and joint defense IT. The message is clear: Europe does not want to remain permanently dependent on Palantir & Co. for critical infrastructure and security software.
This is not just a procurement project, but a sovereignty project. Whoever controls the data, models, and platforms ultimately also controls parts of political agency. Especially in the security sector, that is a sensitive point. Of course, European technology policy often takes a bit longer than a Silicon Valley launch video. But if the result brings more independence, the wait may be worth it. The key will be whether political declarations actually turn into viable products and standards.
📚 Research: Data-Driven Maintenance for Machines
A new paper on arXiv deals with data-driven algorithms for maintaining multiple identical machines under a so-called block replacement policy. Put simply, the goal is to learn the optimal replacement time from operating data instead of setting maintenance intervals rigidly by gut feeling or calendar.
Why is this relevant for AI Radar? Because work like this shows how strongly data-driven optimization is changing classic industrial problems. It may not be a large language model or a flashy consumer launch — but research like this forms the basis for many practical AI applications in production, logistics, and maintenance. For you, that means AI is not just chatbot theater, but also a tool for hard cost optimization. And that is usually less sexy, but much more profitable.
🌏 China Is Building Its Own AI Order as a Counterweight to the West
With a new AI initiative, China is positioning itself even more clearly as a counterweight to the Western-dominated AI system, The Decoder reports. Xi Jinping announced 5,000 AI training places for countries in the Global South at the World AI Conference in Shanghai, along with the founding of the “World Artificial Intelligence Cooperation Organization.” There are also planned cooperation centers with ASEAN, the African Union, and BRICS states.
This is more than symbolism geopolitically. China is actively building networks, standards, and training structures — precisely the building blocks that shape technological spheres of influence. While the West often debates regulation and risk, China is simultaneously investing in reach and partnerships. For companies and observers, this is an important signal: AI is long past being just a product or research competition; it is also foreign policy. Anyone who overlooks that is looking at the models — and missing the game behind them.
🛠️ Tool Tip of the Day
If you want to regularly check your WordPress installation or API surfaces for vulnerabilities, a clean security workflow with vulnerability scans and code review support is worth it. Especially with plugin-heavy setups, it can save you a lot of trouble — and sometimes even a weekend. For getting started with more professional security checks: #
Don’t want to miss any news? Subscribe to the newsletter