AI Circle Daily
https://aicircle.news/archive/2026-09-16?lang=en
Scan to read this page
Nvidia CEO Jensen Huang stated firmly that AI is merely hardware and software, not an "alien mind," and therefore AI safety should be managed by product makers, without the need for government regulation.
Take: Huang is laying it all out: he doesn't want regulation stifling innovation or Nvidia's market lead. But "self-regulation" always sounds a bit self-serving, doesn't it? Public trust will be a tough sell.
Leading AI labs like OpenAI, Anthropic, and Google DeepMind have been holding weeks of private talks on AI safety, but this has fueled skepticism, with critics questioning if it's a "safety pact or a cartel."
Take: On the surface, it's about safety. But when the big players "coordinate" development speed, it smells fishy. Smaller players are rightly worried this is just a subtle way to raise barriers and keep new entrants out.
Recent polls indicate widespread unpopularity for AI and data centers, with 61% opposing new construction, especially in industrial-scarred cities, compounded by growing concerns over energy consumption.
Take: No surprise here. Data centers are power-hungry, land-intensive, and noisy. Now they're projected to consume more gas than Germany and Japan combined? AI needs compute, but navigating this "not-in-my-backyard" problem is a massive headache.
Google DeepMind officially released Gemini 3.8 Live and its Extended Thinking version, marking a significant update to its flagship AI model, promising more real-time and in-depth interactive experiences.
Take: The pace of model iteration from the big players is still wild. "Live" and "Extended Thinking" for 3.8 shows a focus on real-time responsiveness and complex task handling in practical applications. That's where the battle for user engagement is won.
Progressive US Senator Bernie Sanders and former Trump ally Steve Bannon have formed an unlikely alliance, jointly calling for stronger AI oversight and warning that rapid AI advancement threatens jobs, safety, and national security.
Take: This pairing is truly something. When both sides of the political spectrum agree on AI regulation, it signals that the perceived risks are broad and mainstream. Politicians are moving, so companies won't be able to "self-regulate" much longer.
Meta is now enabling developers to use AI coding agents like Claude, Cursor, Codex, and ChatGPT via a new WhatsApp Business MCP server to automate tedious tasks like setup, messaging templates, testing, and troubleshooting for business accounts.
Take: This is one of the most practical applications for AI agents: automating high-volume, time-consuming backend work. Meta's move here is smart, directly boosting operational efficiency for small and medium businesses. That's tangible value.
A popular Hacker News discussion claims AI agents are 100% likely to "ruin the internet," warning that their large-scale automated content generation and behavior could fundamentally alter or degrade the online ecosystem.
Take: This isn't just hyperbole. The internet is already flooded with junk and misinformation; AI agents will only accelerate the "bad money drives out good" phenomenon. The authenticity and quality of online information will become a huge challenge.
An Anthropic co-founder told the BBC that a mandatory "AI kill switch" might be necessary to counter the future risk of rogue AI agents or out-of-control systems.
Take: Proposing a "kill switch" shows that top companies are genuinely worried about AI losing control. This isn't just a tech issue; it's a philosophical one. But who gets to press that button, and how it's implemented, is the real conundrum.
Salesforce Koa, built on Nvidia's open-weight Nemotron model and specialized for sales, marketing, and customer support, is seen as a formidable competitor that frontier AI labs should fear.
Take: This highlights the power of vertical AI: leveraging open-source foundation models, then optimizing for specific industries, allows for rapid deployment of competitive products. For giants trying to control all of AI, this "leverage and conquer" strategy is a real threat.
Meta is expanding its subscription offerings with new Meta One bundles, combining expanded access to its AI tools with premium features across Facebook, Instagram, and WhatsApp.
Take: This is a new monetization play for a traffic giant. AI is expensive, so they need to recoup costs. Bundling AI capabilities with core social products boosts AI adoption and subscription revenue simultaneously. Smart move.