This article first appeared in Digital Edge, The Edge Malaysia Weekly on May 11, 2026 - May 17, 2026

Only two of 24 customer service chatbots tested across Malaysia’s biggest consumer-facing sectors could reliably understand their users, pointing to a comprehension gap between a small group of leaders and the rest.

This is according to a new white paper from artificial intelligence consultancy firm Entermind titled Confessions of an AI Chatbot.

The whitepaper evaluated 24 of the country’s top customer service chatbots across four sectors — e-commerce and e-wallet, travel, telecommunications and financial services — using the Entermind Chatbot Quality Index (CQI).

The CQI is a proprietary framework that scores bots against 26 binary tests grouped into five weighted categories: comprehension, access, experience, functional capability and safeguards.

Only 23% of the chatbots tested correctly understood negation, just 30% could handle a change of topic mid-conversation, and only 33% could recognise key details in a query. Meanwhile, two-thirds were unable to understand Bahasa Melayu, Manglish, or both.

The two chatbots that passed all seven of Entermind’s comprehension tests — covering negation, slang, topic changes and multilingual support — were Touch ‘n Go and Batik Air.

Comprehension proved to be the sharpest differentiator in the index, with 78% of chatbots that passed at least four of the seven comprehension tests also scoring above 60% overall.

Batik Air topped the overall index at 83.3%, scoring a near-perfect profile of 7/7 on comprehension, 3/3 on functional capability, 7/7 on safeguards and 6/6 on experience, with its only weakness a login barrier at the access stage.

This is followed by Touch ‘n Go (82.3%), Firefly (78.1%), RytBank (75%), Maxis (74.2%), Boost (74%) and Malaysia Airlines (72.5%). At the bottom of the table sat TIME (4%), Maybank (20%), U Mobile (20%), UOB (20.8%) and Grab (23%).

The average chatbot scored just 49.5% on the index. Eleven of the 24 scored below 40%, while six did nothing more than greet users and route them to a human agent.

Chatbots scoring above 70% — seven of the 24 tested — shared three common traits: genuine language understanding, end-to-end transaction architecture and robust safety infrastructure.

Another aspect tested in the CQI was functional capability, which was anchored by the ability to perform transactions. Only 14% of chatbots could complete a transaction within the chat itself, with Batik Air, Maxis and Shopee the only three to pass.

Meanwhile, five chatbots achieved a clean 7/7 sweep on Safeguards, covering manipulation resistance, profanity handling and error recovery. These are RytBank, Boost, IndiGo, Batik Air and Malaysia Airlines.

Experience was the weakest category in the index, averaging just 37%. Voice and image input passed at only 16% and multi-format responses at 26%. Even higher-scoring chatbots drew criticism for “wall of text” responses, with evaluators asking them to “get straight to the point”. None of the 24 chatbots tested demonstrated cross-session memory.

On the other hand, access was the only category that was universally strong, averaging above 74%.

The white paper set out six priorities for operators looking to close the gap: investing in language understanding, enabling in-chat transactions, fixing fallback and error handling, designing for experience polish, building cross-session memory and evolving beyond triage models.

Ant International now settles 45% of its cross-border money transfers on-chain using tokenised bank deposits, according to its president Douglas Feagin.

He said the company uses its Whale platform with around 10 global banking partners, including JPMorgan Chase, Citigroup, HSBC and Standard Chartered, enabling round-the-clock transfers at a lower cost than traditional cross-border banking systems.

Feagin added that tokenised deposits offered a more practical route for blockchain-based settlement because they operate within existing banking regulations, avoiding the need for entirely new legal frameworks.

According to him, the Whale platform replaces the traditional multi-bank correspondent model, where transfers pass through several intermediary banks and the SWIFT network before reaching the recipient, adding delays and costs.

Ant International is also exploring broader tokenisation opportunities, including tokenised investments, which Feagin said could allow investors to begin earning interest immediately rather than waiting for settlement periods.

The Whale platform forms part of Ant International’s wider “FinAI” strategy, combining blockchain settlement with AI-driven foreign exchange optimisation and fraud detection tools. — By Danial Dzulkifly

SEEK has launched a new Asia-wide integrated marketing campaign for Jobstreet by SEEK and Jobsdb, placing “relevancy” at the centre of how jobseekers search for work amid growing demand for better job fit rather than more listings.

The campaign, which spans Malaysia, Singapore, Hong Kong, Thailand, Indonesia and the Philippines, builds on SEEK’s “Better Matches” platform and highlights its AI-powered search and matching capabilities through character-led storytelling featuring figures such as the Monkey King and a mermaid.

According to Jobstreet by SEEK data, seven in 10 Malaysians report being happy at work, although nearly one in three remain neutral or unhappy, suggesting many are still seeking roles better aligned with their skills and career goals.

SEEK said its AI-powered tools analyse users’ skills, experience, preferences and behaviour to provide more personalised job recommendations. Features such as the Strong Applicant badge are also aimed at helping candidates identify roles where they are a strong fit.

Ericsson has been ranked the leader in Growth and Innovation in Frost & Sullivan’s Frost Radar: 5G Network Infrastructure, 2026 report for the sixth consecutive year.

The report assessed more than 100 global industry participants, with Ericsson leading across both the growth and innovation axes among the top 20 companies evaluated.

Frost & Sullivan cited Ericsson’s broad 5G portfolio spanning radio access networks, transport, and core and edge networks, alongside its investments in AI-enabled and energy-efficient technologies.

The report also highlighted Ericsson’s research and development spending, which accounted for nearly 21% of revenue in 2025, as well as its progress in reducing energy consumption at radio base station sites.

Syarikat Takaful Malaysia Am Bhd has partnered with Handshakes Technology (Malaysia) Sdn Bhd to deploy a technology-driven compliance and risk monitoring solution.

The collaboration, formalised through a memorandum of understanding, aims to strengthen anti-money laundering and financial crime oversight through continuous, data-driven monitoring aligned with Bank Negara Malaysia’s AML/CFT/CPF and targeted financial sanctions requirements.

Under the partnership, Handshakes will provide ultimate beneficial ownership governance and risk intelligence capabilities, integrating corporate data into onboarding, due diligence and ongoing monitoring processes.

The initiative is intended to improve visibility into ownership structures and support faster, data-led risk assessments across the customer lifecycle.

Handshakes will also extend access to its corporate intelligence platform to selected stakeholders within Takaful Malaysia Am’s network, including bancatakaful partners, intermediaries, loss adjusters and claims solicitors.

More than two-thirds of leaked passwords can now be cracked within a day, according to new research by Kaspersky, highlighting growing risks from predictable password habits and AI-driven cyberattacks.

Analysing 231 million unique passwords leaked between 2023 and 2026, Kaspersky found that 68% could be broken within 24 hours, while over 60% could be cracked in about an hour.

The cybersecurity firm said 53% of compromised passwords ended with digits, while 17% began with them. Common patterns included dates, keyboard sequences such as “1234”, and widely used symbols like “@” and “!”.

Kaspersky also found that internet trends increasingly influenced password choices, with the use of the word “skibidi” surging 36-fold over the period analysed.

The company warned that longer passwords alone no longer guarantee security, as AI-powered cracking tools can break even some 15-character passwords in under a minute if they follow predictable patterns.

On another note, spyware attacks targeting organisations in Southeast Asia rose 18% year on year in 2025. Kaspersky said its business solutions blocked more than 818,000 attacks across the region.

Singapore recorded the sharpest increase at 111%, followed by the Philippines at 85% and Malaysia at 75%. Indonesia saw a 35% rise, while Vietnam logged an 8% increase. Thailand was the only market to record a decline.

Kaspersky warned that threat actors are increasingly using spyware for long-term intelligence gathering rather than disruption alone, targeting sensitive corporate data and internal communications.

In March 2025, the company uncovered “Operation ForumTroll”, a cyberespionage campaign exploiting a Chrome zero-day vulnerability to infiltrate organisations in sectors including media, government, education and finance through phishing emails and spyware tools.

Kaspersky said businesses need intelligence-led and AI-powered cybersecurity solutions to counter increasingly sophisticated attacks.

Digital lending platform Direct Lending has expanded its financing ecosystem by introducing shariah-compliant instalment plans for medical treatments.

The medical treatment instalment plan is designed to assist locals in managing healthcare expenses that may not be fully covered by insurance or for procedures that require immediate funding. The facility covers a broad spectrum of services, ranging from dental to maternity care.

Through partnerships with private clinics and healthcare providers, patients can split bills into monthly instalments of up to 12 months. The move comes amid rising private healthcare costs and reflects Direct Lending’s focus on providing targeted financing solutions for underserved consumers.

Financial institutions are creating new artificial intelligence (AI) governance risks faster than they can manage existing ones, according to Netskope’s Threat Labs Report: Financial Services 2026.

The report found that while the proportion of employees using personal AI accounts fell from 76% to 36% over the past year, the number switching between personal and corporate AI accounts rose from 9% to 15%.

It also revealed that 97% of financial services employees now use AI indirectly through embedded features in workplace productivity software, increasing the difficulty of monitoring and enforcing security controls.

Regulated financial and customer data accounted for 59% of all recorded data policy violations on generative AI platforms and 65% on personal cloud applications last year.

Despite restrictions on certain AI tools, adoption remains widespread. OpenAI’s ChatGPT is used in 76% of financial services organisations, while Google’s Gemini is used in 68%. Meanwhile, AssemblyAI’s adoption rose from 1% to 37% within a year.

The report also warned that attackers are increasingly exploiting trusted cloud platforms to distribute malware. GitHub was identified as the leading malware delivery channel, affecting 11% of organisations, followed by Microsoft OneDrive at 8.2%.

Netskope said financial institutions need broader monitoring of web and cloud traffic and stronger data controls tailored for AI tools and personal applications.

If the “Lofi Girl” YouTube stream has been your unofficial co-worker for years, it is time for a digital promotion. Tokyo-based developer Nestopi Inc has officially brought its critically acclaimed productivity game, Chill with You: Lo-Fi Story, to mobile devices, offering a cosy sanctuary for those of us who struggle to stay on task in a world of distractions.

Far from being just another music player, this “productivity adventure” allows you to co-work alongside Satone, an aspiring novelist. The concept leverages the psychological benefits of “body doubling” — the idea that working alongside someone else can boost focus — but replaces the awkwardness of a real-life office with soft scenery, ambient rain, and a curated library of over 30 lo-fi tracks.

Having already secured an “overwhelmingly positive” rating on Steam with over 9,600 reviews, the transition to iOS and Android is a logical step for the work-from-cafe crowd. Key features include cross-platform sync, where users can start their to-do list on their PC and sync it seamlessly to their smartphone for their commute, and narrative growth, which rewards users’ deep work sessions by unfolding a story between the user and Satone.

There are also custom environments where users can adjust ambient sounds and visuals to match their specific vibe for the day.

While the mobile version is free to download — offering the first 10 story episodes and basic focus tools — serious grinders can upgrade to the full version.

A 130cm-tall humanoid robot, a Unitree G1 model, was ordained as a “robot monk” in South Korea on May 6 by the Jogye Order of Korean Buddhism at Jogyesa Temple. The robot, named Gabi — the Korean word for mercy — was introduced to symbolise the coexistence of technology and religion. During the ordination ceremony, monks asked the robot whether it would devote itself to Buddha and Buddhist teachings, to which it replied: “Yes, I will devote myself.” The robot was also given five adapted Buddhist precepts: respecting life, not harming robots or objects, obeying humans, avoiding deception and conserving energy. “At the current stage, it might seem a bit awkward ... but we hope the five precepts will be considered as basic principles for not only Buddhists but the society to exist together with robots,” Venerable Seong Won, who oversees cultural affairs at the Jogye Order, was quoted saying in The Korea Times.

In an era where smartphone screens are getting brighter and more power-hungry, Oasis Phoenix’s InkCase Plus offers a refreshing, low-energy alternative. This second-screen accessory for Android devices features a 3.5-inch E Ink display, bringing the power of electronic paper to the palm of your hand without the bulk of a traditional e-reader.

The primary appeal lies in its sunlight-readable, low-power consumption. By offloading static tasks — such as reading e-books, viewing boarding passes or checking notifications — to the E Ink screen, users can significantly extend their phone’s battery life. Unlike traditional LCDs, the E Ink panel only consumes power when the image changes, allowing your favourite photos or essential to-do lists to remain visible indefinitely without draining resources.

The InkCase Plus can be embedded into a dedicated FitCase or used as a standalone display. This flexibility is ideal for professionals who need constant access to information without the constant glare of a backlit screen. The device also supports specialised apps, enabling users to push news headlines, weather updates and fitness tracking data directly to the secondary display via Bluetooth.

For those looking to bridge the gap between digital convenience and paper-like comfort, the InkCase Plus is a compelling addition to any tech arsenal. It’s priced at RM162 on www.inkcase.com/inkcase-pix, with occasional discounts available. 

The AI Ideal: Aidealism and the Governance of AI offers an actionable vision for ensuring AI strengthens democracy, ethics and human dignity. Instead of allowing AI to concentrate power in the hands of a few, the book argues for a new global framework  one where AI serves justice, enlightenment and human betterment.

Rooted in European Enlightenment ideals, Scandinavian social models and liberalism, and Swiss direct democracy, Aidealism rejects extreme ideologies and champions pragmatic, ethical and forward-thinking solutions. From free education and healthcare to AI-driven economic justice and climate responsibility, this book explores how AI can help build a sustainable, free and prosperous world.

Instead of a warning of the catastrophe of AI, Lidströmer offers an actionable vision for ensuring AI strengthens democracy, ethics and human dignity. The book explicitly gives a manifesto for practical action, including a plan for how to harness and use AI for the common good so that it benefits everyone, not just the few. It elaborates on the daily conundrums of the human species — our nature, origins, goodness and cruelty, memes, hierarchies, political structures, and how to build a fairer, more just, peaceful and benevolent society.

As the risks are real and the threats are mounting, sections cover how AI could empower autocrats, disrupt economies and undermine human agency while also highlighting how AI could also be our greatest tool for wisdom, fairness and progress — if governed with foresight and courage. — Amazon

Save by subscribing to us for your print and/or digital copy.

P/S: The Edge is also available on Apple's App Store and Android's Google Play.