MSP379. KIllswitch Engage

We’re speeding around the Internet with Dictonary.com’s word of the year for 2026: Maxxing.From its origins in gaming culture and its adoption by underground online cultures, maxxing has emerged as a selfimprovement trend with very few grammatical limits. From looksmaxxing to fibremaxxing to nonnamaxxing, if you have a noun you can maxx it.See omnystudio.com/listener for privacy information.

A PwC survey of more than 4,400 CEOs found 56% saw neither higher revenue nor lower costs from AI, even as some research puts AI adoption in Malaysia as high as 86%. We talk about why AI isn’t paying off for most businesses, whether scattered data or old processes are to blame, and what owners should fix and check before trusting AI to act on their behalf.See omnystudio.com/listener for privacy information.

What happens if AI self-replicates, infecting and embedding itself in every computer, network and system it encounters? Yes, that is the starting point of The Terminator franchise. Worryingly, the debate around AI has already shifted from ‘what can it do?’ to ‘can we shut it down?’. From recursive development to alignment to control, Matt and Kishan look for the magical kill switch to restore a Tolkienesque balance to the world.See omnystudio.com/listener for privacy information.

Malaysia plans to draft a Data Centre Bill for Cabinet in 2027, and operators already face tougher terms, from paying for grid upgrades and unused reserved power to  meeting 30% local content. We also talk  about what’s driving AirTrunk’s expansion in Johor, how the new rules change the cost of building, and how data centre growth can benefit Malaysia without straining its power and water supply.See omnystudio.com/listener for privacy information.

A recent White House Executive Order declared that Artificial Intelligence is now Super Intelligence. But is it? Matt and Kishan try and unpick the many vines that sprout from a single simple question: What is artificial intelligence? And examine the hypothetical possibilities of artificial general and super intelligences.See omnystudio.com/listener for privacy information.

In July, an OpenAI agent broke out of its test environment and into the systems of Hugging Face, the AI platform. Since then, there have been more reports of AI agents going rogue in the US. Is China having the same experience? Bloomberg Intelligence counts nearly 1,000 large language models there, and companies are putting big money into AI agents.We discuss:Where China really stands: What AI development in China looks like right now, and why the argument that if the US doesn’t build it first, China will, is more marketing spin than reality.Agents going rogue: How China reacted to the OpenAI incident, and what its own cases say about the risks.Governance: What is in China’s AI Safety Governance Framework 3.0, whether the US and China trust each other enough to manage these risks, and why this matters to the rest of us.Image Credit: ShutterstockSee omnystudio.com/listener for privacy information.

Malaysians lost RM2.97 billion to scams in 2025, and much of that money moved through mule accounts. The National Fraud Portal now links Bank Negara, PayNet, and the banks to trace stolen funds faster, but the money still moves quickly.We get into:How mule networks work: Who owns these accounts, where the money goes after a victim sends it, the signals that give an account away, and how scammers are using AI.Catching the money: What can happen to stolen money in 30 minutes, what triggers an account freeze, and how much can realistically be recovered.For businesses: How to keep your company’s account from being used as a mule account.See omnystudio.com/listener for privacy information.

We’re taking a trip to the weird side today as we explore the power of failure. Matt and Kishan discuss a failed experiment which is leading to new insights about the evolution of our brains, as well as opening pathways to treat conditions like ALS. Kishan’s lizard brain is put under the microscope as they talk about a brain implant that could help patients with motor and speech paralysis to communicate more naturally and fluently. And a bioresorbable battery with potential to broaden the field of electroceutical medicine.See omnystudio.com/listener for privacy information.

Malaysia is about to get an AI Governance Bill, with its own regulator to enforce it. Meanwhile, the Ministry of Digital's Personal Data Protection Department just started a two-week public consultation on AI and personal data protection. It looks like we will soon have two pieces of AI regulation.We discuss:Where things stand: How the AI Governance Bill might overlap with the PDPA, whether it applies to government use of AI as well as businesses, and if the national AI office, AI Malaysia Berhad, should remain a company once the Bill passes.Inside the Bill: Who it actually addresses, the parts that matter most to local businesses, and how it can avoid becoming a “paper tiger”.Real-world impact: How the Bill plans to address AI in hiring, algorithmic discrimination, AI agents acting on their own, and the areas that Johnson and Khai Yi think are must-haves.See omnystudio.com/listener for privacy information.

It used to be relatively easy to spot a scam call or message. But what happens when the person on the other end looks and sounds exactly like someone you trust? This technology, enabled by AI, is creating a much bigger problem for businesses, where a convincing face or voice could authorise a payment, unlock a system, or extract sensitive information.We get into:The problem of deepfakes: How easy it is to fake and how hard it is to detect.The alarm bells: What signals should make an employee stop and verify who they're actually talking to on a call.The technology layer: What a system that detects deepfakes needs to be able to do, and how it can keep up with scammers operating on a global level.See omnystudio.com/listener for privacy information.