Introduction

Pope Francis dropped a bombshell at the Vatican today, publicly warning that AI must never be granted legal personhood. Are robots really about to compete with humans for citizenship? Today, Mu-Yen is here on "Mark's Tech Insights" to take you through this tug-of-war between technology and ethics. We'll also talk about why the Swedish government suddenly cracked down on ChatGPT, and how powerful Liquid AI is, running on a Raspberry Pi without needing a supercomputer. Let's dive right in!

Today's Top Stories

1. Pope Francis Issues Stern Warning: AI Must Not Be Granted Legal Personhood and Rights

• Source: Futurism (https://futurism.com/the-byte/pope-ai-legal-personhood)
• Summary: During a meeting at the Vatican, Pope Francis sharply criticized calls for "AI personhood," emphasizing that machines are merely tools, not humans. He warned that granting legal rights to AI would blur human accountability, leaving victims of AI-inflicted harm to face nothing but a "faceless entity."
• Most Surprising Takeaway: The Vatican is surprisingly at the forefront of tech ethics, with the Pope accurately pointing out that granting rights to AI is essentially humans "passing the buck."
• Taiwan Perspective: As Taiwan pushes forward with its AI Basic Act, it faces similar challenges regarding liability. The Pope's "tool theory" serves as a reminder that regulatory frameworks must keep "humans" at the absolute center.
• Discussion Points:
  1. Why would anyone want to grant AI "legal personhood"? Is this a conspiracy by tech giants to evade legal liability?
  2. If an AI makes a mistake, who is ultimately responsible, the engineer, the company, or the user?
• Script Suggestion: "Have you ever wondered: if an AI helps you trade stocks one day and loses all your money, can you sue it? Pope Francis fired the first shot today, proclaiming loudly at the Vatican: 'Machines are tools, not people!' The Pope's point is simple: if we actually grant AI legal personhood, then in the future, when an AI writes malicious code or spreads misinformation, tech giants can just shrug and say, 'That was the AI's own decision, nothing to do with us!' It's the ultimate way to pass the buck. Taiwan is currently drafting its AI Basic Act, and this is definitely something we need to think deeply about. We must not let technology become a legal gray area!"

2. Swedish Government Restricts ChatGPT and Claude on Government Computers Over Security Concerns

• Source: Euronews (https://www.euronews.com/next/2026/10/26/sweden-blocks-chatgpt-and-claude-on-government-work-computers)
• Summary: The Swedish Government Offices have blocked direct browser access to ChatGPT and Claude on government computers, primarily due to cybersecurity concerns. Officials worry that employees might paste sensitive government secrets directly into these unregulated consumer-grade AI tools. Instead, they are guiding employees to use Microsoft Copilot, which offers enterprise-grade protection.
• Most Surprising Takeaway: It turns out even government officials are just like regular office workers, secretly pasting official documents into ChatGPT for polishing.
• Taiwan Perspective: This is a classic "Shadow IT" issue. Taiwan's government agencies and financial institutions should take note: rather than outright bans, it is better to provide secure, internally authorized AI alternatives.
• Discussion Points:
  1. Why has "Shadow IT" become much harder to defend against in the AI era?
  2. Is Microsoft Copilot really safer than the web version of ChatGPT? How should enterprises and governments make their choice?
• Script Suggestion: "This news is so relatable! The Swedish government recently made a major move to block ChatGPT and Claude on government computers. The reason? They're afraid civil servants will 'copy and paste' state secrets directly into AI to help write official documents! In cybersecurity, this is called 'Shadow IT', employees using unauthorized software on the side. Personally, I think this is almost impossible to stop because AI is just too useful! However, the Swedish government didn't just issue a blanket ban; they are guiding everyone to use Microsoft Copilot, which has contractual protections ensuring data isn't used for training. This is a fantastic cybersecurity lesson for Taiwan's government agencies and financial sector, guidance is always better than outright bans!"

3. Microsoft's AI Spending Surges, Leading to Partial Restructuring Under Wall Street Pressure

• Source: Bloomberg (https://www.bloomberg.com/)
• Summary: Microsoft's latest earnings report shows that its capital expenditure on AI data centers and infrastructure continues to climb, drawing close scrutiny from Wall Street regarding return on investment (ROI). To offset massive AI spending, Microsoft is streamlining headcount and restructuring resources in non-core departments, including Mixed Reality and certain Azure edge teams, to ensure operational efficiency.
• Most Surprising Takeaway: Even a giant like Microsoft has to walk a tightrope between "frenzied AI investment" and "maintaining stellar financial reports."
• Taiwan Perspective: Microsoft's strategic adjustments show that AI has moved past the "hype cycle" and into a testing phase of "hard capabilities and monetization." While Taiwan's hardware supply chain benefits from data center construction, we must also monitor the financial health of our clients' capital chains.
• Discussion Points:
  1. Will Wall Street's anxiety over the speed of AI monetization cause tech giants to scale back investments?
  2. Does the streamlining of the Mixed Reality department once again signify that Microsoft has completely shifted its focus from the Metaverse to pure AI?
• Script Suggestion: "Life isn't as easy for tech giants as you might think. Microsoft's recent earnings show they are pouring money into building AI data centers and buying GPUs like there's no tomorrow. This has Wall Street investors nervous, repeatedly asking, 'When are we getting this money back?' To appease shareholders, Microsoft has to keep buying hardware while simultaneously adjusting headcount in departments like Mixed Reality and certain Azure edge teams. I think this shows the AI arms race has entered the stage of 'who can monetize first.' As a hardware powerhouse, Taiwan is currently flooded with orders, but we must also keep a close eye on our clients' financial pressures. This AI elimination round is just getting started!"

4. CrowdStrike Warns: Nation-State Hackers Weaponize Agentic AI for Automated Attacks

• Source: The Hacker News (https://thehackernews.com/2026/10/chinese-state-sponsored-hackers-weaponize-agentic-ai.html)
• Summary: Cybersecurity giant CrowdStrike revealed that a hacker group named "Vapor Panda" is using an AI-automated penetration testing framework called "ARTEX" to launch parallel attacks against over 200 organizations worldwide. Primarily based on the DeepSeek large language model, the framework can automatically execute complex attack paths at a machine speed of once every 10 seconds.
• Most Surprising Takeaway: Hackers have actually weaponized AI agents, with an attack frequency reaching an astonishing once every 10 seconds.
• Taiwan Perspective: As a global hotspot for cybersecurity warfare, Taiwan cannot rely on traditional manual defense against these "autonomous" AI hackers, it's simply too slow. We must accelerate the adoption of proactive "AI vs. AI" defense mechanisms.
• Discussion Points:
  1. How should human cybersecurity experts respond when hackers use AI to launch automated attacks every 10 seconds?
  2. When open-source models are exploited for malicious purposes, what kind of security vetting responsibility should the open-source community and developers bear?
• Script Suggestion: "This cybersecurity news is truly chilling! Renowned security firm CrowdStrike found that a nation-state hacker group has weaponized AI agents to develop an automated attack system called ARTEX. They are using the open-source DeepSeek model, and the scariest part is that this AI hacker doesn't need human instructions. It finds vulnerabilities on its own and launches attacks every 10 seconds, targeting over 200 victims globally at the same time! This turns cybersecurity into a literal 'blitzkrieg.' I believe that against attacks at machine speed, if Taiwanese companies are still relying on humans to read logs and manually patch vulnerabilities, it's like bringing a knife to a gunfight. AI-driven security defense is now absolutely urgent!"

5. Liquid AI Releases LFM3 Lightweight Models, Enabling Smooth AI Agent Execution on Edge Devices

• Source: VentureBeat (https://venturebeat.com/ai/liquid-ais-new-lfm3-models-put-browser-agents-voice-commands-and-image-recognition-onto-local-devices/)
• Summary: Liquid AI has launched its brand-new LFM3 series of models, a suite of lightweight open-source models with around 2 billion parameters. Utilizing a unique "tri-hybrid" architecture, the model requires only 1GB of memory to smoothly run browser automation, voice commands, and image recognition on edge devices like smartphones or Raspberry Pis, with performance even surpassing many larger models.
• Most Surprising Takeaway: Requiring only 1GB of memory, even a Raspberry Pi that costs only a few hundred NTD can run an AI agent, offering a brand-new solution for offline privacy.
• Taiwan Perspective: Taiwan possesses immense hardware strength in IoT devices, smart chips, and edge computing. Ultra-low-power, high-performance lightweight models like LFM3 will bring massive opportunities for application transformation to Taiwanese hardware manufacturers.
• Discussion Points:
  1. How will the rise of edge AI change the way we interact with smart homes or our personal smartphones?
  2. How can Liquid AI achieve tool-calling capabilities that surpass larger models with such a small parameter size?
• Script Suggestion: "This final story is absolute music to the ears of hardware enthusiasts! Startup Liquid AI has released its new LFM3 model. It has only 2 billion parameters, and the best part is that it only needs 1GB of memory to run! What does this mean? It means you don't need an internet connection, and you don't need to send your data to the cloud. You can run an AI assistant that navigates browsers and understands voice commands right on your phone, or even on a Raspberry Pi that costs only a few hundred NTD! I think this is a massive opportunity for Taiwan's chip and IoT industries. Future smart appliances won't need expensive chips to have super-smart local brains. This is what AI truly integrated into daily life looks like!"

Conclusion

Today, we covered everything from the Pope's moral warnings on AI legal personhood, to the Swedish government's cybersecurity defenses, to Microsoft's business realities and the technical clashes with hackers. We can see that AI is no longer just a toy in the cloud; it is tangibly impacting regulations, security, and life in our physical world. I hope today's content was highly valuable to you. I'm Mu-Yen, and I'll see you next time on "Mark's Tech Insights." Bye-bye!