Singapore AI News & Daily Briefing

Singapore is launching a global institute focused on AI governance specifically for the financial sector. The institute will help banks and financial firms adopt AI responsibly, ensuring compliance with regulations and managing risks like bias and data privacy. For everyday users, their banking apps and financial advice may become more AI-powered but also better supervised.

OpenAI is giving ChatGPT its biggest update yet, turning it into a 'superapp' that combines coding tools, image generation, and partner services like Canva and Booking.com. The overhaul aims to boost revenue ahead of a potential stock market listing. For users, ChatGPT will become more of an all-in-one assistant, especially for coding tasks.

French AI lab Mistral is reportedly in talks to raise €3 billion at a €20 billion valuation, nearly doubling its previous €11.7 billion valuation from September. One of Europe's leading AI startups, Mistral has positioned itself as a 'sovereign' alternative to American AI firms, partnering with France's army, Luxembourg's government, and major European companies. Despite the raise, Mistral has only raised about $4 billion to date—a fraction of what OpenAI ($186B) and Anthropic ($161B) have taken in.

Google has sued a Chinese cybercrime group called 'Outsider Enterprise' for using AI to orchestrate mass text message scams. The group sent 2.5 million scam texts over just two weeks, targeting hundreds of thousands of victims. Google is seeking legal action against the operation that used AI to scale up fraud.

Amazon's cybersecurity team reportedly flagged Anthropic's Fable and Mythos AI models as potential national security risks, which helped convince the White House to issue an export control order. This led Anthropic to cut off access to these models. The move shows how private tech companies can influence government AI policy — and how seriously Washington is treating the risks of advanced AI systems getting into the wrong hands.

Tech in Asia compiled a list of 50 AI startups in Asia that recently raised funding, covering up to Series E. The list is regularly updated and available to subscribers, giving a snapshot of where investment is flowing in Asian AI.

Chinese AI chip manufacturers are rushing to go public after recent IPOs showed massive investor appetite. Moore Threads Technology saw its stock jump 425% on debut, while MetaX Integrated Circuits gained 693% on their first trading day. Shanghai Biren Technology is now planning a Hong Kong listing to raise up to US$600 million. This reflects China's push for technological self-reliance in AI hardware amid US export restrictions.

As Anthropic suspended its newest AI models under US government orders, India's tech leaders are using the episode to push for stronger domestic AI capabilities. The incident has sparked debate about whether India should rely on American AI systems or invest more in homegrown alternatives — a question many countries are quietly asking themselves right now.

A PwC report reveals that AI tools in hospitals are being used to maximize billing—essentially helping doctors code patient visits more complexly so hospitals get paid more per encounter. The result? Medical costs are projected to rise nearly 9% next year, and AI is partly to blame. Rather than reducing healthcare costs, AI documentation tools are capturing more billable details that translate to higher bills for patients and insurers.

Anthropic released its most powerful public AI model, Claude Fable 5, on June 9. Within three days, a US government export control directive blocked global access to it. This marks one of the fastest government crackdowns on an AI model ever, raising serious questions about how freely countries can access frontier AI technology and what happens to companies building the world's most advanced systems when geopolitics intervenes.

Two AI note-taking tools tested in real Singapore settings: the $179 Plaud Note Pro (hardware device for capturing meetings and group conversations) and Google Eloquent (free iPhone app for personal dictation). Plaud excels at picking up voices across a room; Eloquent turns your own speech into clean, shareable text. Both stumble slightly on Singaporean accents but remain useful for professionals who need reliable meeting notes.

Amazon is pouring over $33 billion into cloud and AI infrastructure across Southeast Asia — Indonesia, Malaysia, Singapore, and Thailand — by 2039. The company says this will create more than 56,300 full-time jobs each year and add over $64 billion to the region's GDP. Amazon has already invested more than $3 billion in the region in 2025 alone.

Iranian state media reported a limited cyberattack disrupted services at four banks. The scope and perpetrators remain unclear, but bank disruptions signal potential impact on everyday financial services.

Communities across America are fighting back against massive AI data centers being built near their homes. Residents are worried about noise, water usage, electricity costs, and environmental damage. Local governments are responding with construction bans and moratoriums. Over 1,400 data centers have already been built or approved across 45 states.

Amazon CEO Andy Jassy and five other tech executives reportedly warned the Trump administration about security flaws in Anthropic's Fable AI model — even though Amazon is one of Anthropic's biggest investors. Within hours, the White House forced the model offline via an export control order. This reveals how even investors can push government crackdowns on AI rivals, raising questions about whether security concerns are genuine or convenient excuses.

Dario Amodei, CEO of Anthropic, is proposing that AI companies be taxed to fund universal basic income for workers displaced by automation. He argues that since AI companies benefit from labor being automated away, they should help support people who lose their jobs. The idea is gaining traction as a way to address growing concerns about AI replacing white-collar roles.

AI coding tools like Claude Code can reliably find the right file to edit but often miss the exact lines that need fixing. A new benchmark called SWE-Explore tested this separately from actual code repair and found that without enough context, even the best fix will fail. For developers relying on these tools, this means you might think the AI found the bug when it actually hasn't.

Honor, the Chinese smartphone brand, debuted a 'robot phone' developed with film equipment maker Arri at the Shanghai Film Festival red carpet. Details are sparse, but the collaboration suggests a device combining Honor's mobile technology with Arri's cinema-grade optics expertise.

China's biggest electric car maker BYD is secretly building humanoid robots under a project called 'Yao-Shun-Yu,' aiming to create machines with both strong brains and strong bodies—something US robots supposedly lack. BYD plans to deploy the robots as store greeters and factory workers at its own facilities. With Tesla already mass-producing its Optimus robot and XPeng joining the race, the line between car companies and robotics companies is blurring fast.

KPMG published a report promoting AI adoption to business clients—but the case studies were fake. Made-up success stories involving real organizations like UBS and the NHS were exposed by AI detection experts. KPMG has pulled the report. This raises red flags about how much businesses can trust consulting firms pushing AI solutions.

China is eliminating 12,000 university degree programs deemed 'obsolete' as the country races to retrain workers for the AI era. Universities are shutting down programs in traditional fields and launching new ones focused on AI skills, semiconductor technology, and robotics. This represents one of the largest educational overhauls in history, directly tied to fears that AI will automate millions of jobs.

A year after Meta spent $14.3 billion to hire Alexandr Wang (former CEO of Scale AI) to rebuild its AI capabilities, Meta has delivered its first proprietary AI model called Muse Spark. Now comes the hard part: making money from it. Meta still relies on advertising for 98% of revenue, and Wall Street is skeptical that AI subscriptions or developer tools will become significant income sources. The stock has dropped 18% despite strong revenue growth elsewhere.

Kyndryl, a major IT services company, launched a new AI orchestration tool to help businesses move AI 'agents' from experimental projects into actual daily operations. Think of it like a traffic controller for AI—coordinating multiple AI tools so they work together smoothly across a company. The focus is on industries like retail and logistics where AI could handle tasks like supply chain adjustments or customer service automatically.

Microsoft Research developed Mirage, an AI video generation system that maintains spatial consistency across long camera movements by storing scene information in 'latent space' rather than traditional point clouds. This approach is faster and uses less memory while keeping scenes coherent. The catch: it still struggles to track moving objects reliably across video segments.

Google Cloud launched the Open Knowledge Format (OKF), a new standard that converts scattered company documents into organized Markdown files with structured metadata. This makes internal knowledge usable for AI agents. It's essentially a formalized version of the 'LLM Wiki' concept popularized by AI researcher Andrej Karpathy, and it could become a common way for businesses to prepare data for AI systems.

Financial institutions are navigating the tricky path of AI adoption, with common pitfalls including biased decision-making, poor data quality, and lack of transparency. The piece offers practical guidance on what banks should avoid when deploying AI systems.

A premium $125,000 robot companion is being marketed as a social device, though early impressions suggest it raises more questions than answers about whether high-end robotics can genuinely serve as everyday companions.

A humanoid robot is being prepared to climb Mount Everest, demonstrating advanced robotics and AI capabilities in extreme environments.

Anthropic staff are scheduled to meet with White House officials next week, according to Axios. This follows the US government's recent order blocking foreign access to Anthropic's most advanced AI models, raising questions about the company's regulatory future.

Canadian Prime Minister Mark Carney says US restrictions on AI chips and model access highlight the dangers of relying too heavily on American technology. This comes as the US has been tightening controls on advanced AI exports. For everyday tech users, such moves could eventually affect which AI tools are available and how much they cost.

The US government restricted exports of Anthropic's Mythos AI after intelligence suggested China-linked hackers may have accessed it. This is a follow-up to the recent government shutdown of Anthropic's advanced models. The concern: if adversarial nations have access to frontier AI models, they could exploit them for cyberattacks, disinformation, or military applications. It signals that AI models are now treated like strategic weapons subject to export controls.

Kaspersky spotted scammers abusing Tencent EdgeOne Pages, a legitimate AI web-building platform, to trick workers into handing over corporate login details. Over 8,000 phishing emails were caught in 30 days, aimed at industrial companies and written in English, Korean, and Russian. The hitch: the platform looks trustworthy, so victims are more likely to click.