Review: ATxSummit (Singapore)

How to build a better future with AI safety and Quantum Satellite. And more.

May 29, 2025

Last week, I was invited to attend a conference in Singapore. It’s interesting to know that its agenda resonated with what I have learned and shared with you in the past one year. Today, I would like to give you a quick review of the 2 day summit with some highlights. You can learn it by yourself at https://atxsummitvp.com/ Cheers, Emma.

Singapore has long been admired for doing things right—and this year's AI Summit confirmed it again. The city-state welcomed thought leaders, researchers, technologists, and policymakers from around the globe to discuss one question: how can we use AI to build a better future? What followed was a series of deep, sometimes emotional, always meaningful conversations.

Here’s a simple breakdown of what stood out the most—from AI safety to quantum computing to reach the new sky limit.

1. AI for Social Good: From Drug Discovery to Protecting Children

One of the main themes of the summit was how AI can accelerate social good. We’re seeing real impact already:

Drug discovery is moving faster thanks to AI models that can simulate how molecules interact.
Healthcare is becoming more accessible, especially in regions with limited medical professionals.
Language barriers are being broken down, opening up services and information to more people. Very important for micro, small, medium businesses.

But it’s not all positive. There was a sobering reminder that we have failed to protect our kids from online dangers—addiction, predators, misinformation. As one speaker put it: "We failed to protect the most vulnerable. Let’s not fail again."

That message rang clear across multiple panels: technology is only as good as the people and systems guiding it.

2. AI Safety Is Not Optional

Safety was a hot topic in nearly every session. It’s not enough for AI systems to work—they need to be safe, reliable, and explainable in the real world.

A few key takeaways:

Meta is creating new spaces tailored to teens on Instagram, emphasizing safety.
Adobe is working on a content authenticity initiative to fight deepfakes and misinformation.
There was a strong call for "safety by design": systems that are secure from the ground up, not patched after the fact.

An important insight came from the Global AI Assurance Pilot, which aims to test AI systems the way we test hospitals, airports, and banks—rigorously and repeatedly. Think real-world testing, human-in-the-loop designs, and security by default.

3. Agentic AI: The Next Evolution of Enterprise

Agentic AI, or AI systems that can act independently to carry out complex tasks, was one of the most exciting (and contentious) topics.

Grab said we’re still early; scaling AI agents will need more maturity, data governance, and trust.
Paypal is already imagining a future where agentic AI handles entire transaction workflows.
Visa sees a future where humans and AI work together, not one replacing the other.

For now, these agents are like AI assistants. But in the near future? They might be doing everything from customer service to cybersecurity, all on their own.

4. Mistral AI: A Fresh Take on Open-Source and Safety

Arthur Mensch from Mistral AI offered a grounded, thoughtful perspective on building responsible AI. His key points:

Open source is powerful: It offers low cost, fast deployment, and low latency.
Safety must be built in: Mistral follows a "zero trust" model—never trust user input blindly.
The key is building filters, evaluations, and safeguards into every system.

His core idea? AI is just software. A single system that solves every problems. AGI is a product of problem. Intelligence to solve many cases.

5. Quantum Computing: From Theory to Enterprise

Quantum was another hot trend. It’s no longer just science fiction—it’s becoming a serious tech domain that companies and governments are pouring billions into.

Microsoft shared how it’s spent 17 years building up its quantum team.
Quantum AI systems that can outperform classical systems in areas like drug discovery and vaccine design. (Plus: it’s already real).
Instead of "just another tech," quantum is seen as zero-to-one—a full leap into the future. It’s an entire Apollo program, not a ladder, to the moon.
Need government support for funding and co-design policy that shapes society.
5-6 years to see real things. We are at phase transition for large scale application, performance enhancement.

Singapore is positioning itself as a prime hub for quantum research. Its support system, funding landscape, and tech-savvy population make it an ideal location to launch real-world quantum systems.

6. Robots, Cameras, and Common Sense

The session on embodied AI (robots that interact with the real world) was eye-opening. While robots are improving, they still lack what children gain quickly: common sense.

Some key ideas:

70–80% of safety testing happens in simulations before real-world deployment.
Context matters: robots need to recognize unusual environments, adapt, and communicate.
Human-robot interaction is critical—not just function, but trust, awareness, and understanding.

Google DeepMind shared four principles for robot safety, emphasizing that robots must be designed with the same care as aircraft or medical devices.

7. The Space Economy: A New Frontier (???)

Beyond the Earth, speakers painted a hopeful picture of space as the next economic engine.

Think quantum satellite, super high speed internet and entirely new industries built around them.
Access demands by more countries, lower barrier for prices and benefit of high speed internet everywhere.
Regulations remains as challenges. Partnership is a key to make things happens fast.

Some even argued that this bold, forward-thinking industry could pull us out of the space potential conflicts —a clear call for innovation over regulation.

8. What’s Next for AI?

Toward the end of the summit, attention turned to the future:

AI that cures disease.
AI that finds new energy sources.
AI that makes the world safer—not more dangerous.
And AI that helps us dream again—from roads to skies.

It’s not about sci-fi anymore. GenAI has blurred the lines between fiction and reality. We’re already seeing how fast change can come—and we need to be ready.

Thanks for reading Secure GenAI ! This post is public so feel free to share it.

Final Thoughts: What Singapore Taught Us

There’s a reason why so many people look to Singapore: it’s a place where vision meets action. The summit wasn’t just about talking—it was about doing. Testing. Piloting. Building.

What’s more: