Asia Pacific leaders warn AI adoption brings fresh risks

Technology and security leaders across Asia Pacific are using AI Appreciation Day to highlight both the rapid progress of artificial intelligence and the risks emerging as it spreads deeper into business operations.

The comments reflect a region moving from experimentation to large-scale AI deployment while grappling with workforce disruption, data exposure, and rising cyber threats.

Workplace researchers point to a growing gap between how organisations talk about AI and how employees experience it. Dr Rebecca Hinds, Head of the Work AI Institute at Glean, said many workers are already feeling the effects of automation within their roles rather than through outright job losses.

"The debate over whether AI will replace jobs is still too binary. The most immediate change is happening inside jobs, as AI takes over individual tasks and changes how people learn, contribute, and demonstrate their value. Forty-eight percent of Australian digital workers fear AI could eliminate their role, while 58% say it has already automated meaningful work they would have preferred to keep. But the human work has not disappeared. Employees are still supplying the context, checking the outputs, correcting mistakes, and taking responsibility when AI gets something wrong. The real risk is not simply that AI replaces people. It's that organisations automate the parts of work through which people develop judgement, expertise, and ownership, while leaving them with the clean-up. Leaders should stop asking how many roles AI can remove and start asking how work should be redesigned - what AI should do, what humans must continue to own, and how the productivity gains will be reinvested in people. Companies that use AI only as a headcount lever may cut costs in the short term, but they will also hollow out the skills and institutional knowledge they need to compete," said Dr Rebecca Hinds, Head of the Work AI Institute at Glean.

Cybersecurity executives describe a parallel shift within security teams, where AI agents and automation now handle work that once consumed entire analyst shifts. Kash Sharma, Managing Director Australia and New Zealand at BlueVoyant, said new autonomous tools are reshaping both productivity and risk.

"Across APAC, AI has moved beyond something people prompt to something that acts on its own. A growing list of third-party agents are reading data, calling other systems, and completing multi-step tasks on behalf of staff, often faster and more consistently than a person managing the same workload manually. Security teams are seeing genuine gains too: alert triage that used to take analysts hours now happens in minutes, and case summaries that once took a whole shift can be ready before an analyst has finished their coffee.

"However, most businesses are sitting on years of 'security debt' - files shared too widely, permissions nobody has reviewed, and accounts with more access than they need. An agent will happily draw on all of it when answering a question. The good news is this is one of the most solvable problems in security. Reviewing who has access to what, labelling sensitive data so protection travels with it, and holding agents to the same identity rules as staff are all well-established practices, not new inventions. It's housekeeping, and the payoff is being able to hand agents bigger jobs with confidence.

"The next 12 months will decide who gets the most out of this shift. AI agents will take on bigger jobs, working together across entire workflows rather than single tasks, and the region is well placed to lead. APAC businesses have adopted this technology faster than most of the world, and those that match that speed with proper controls will set the global standard for secure AI adoption. That's the opportunity worth talking about this AI Appreciation Day: not just what AI can do, but proving it can be done right," said Sharma.

Security concerns extend beyond agents into the wider AI stack. Jack Wang, Senior Director, ASEAN, Korea and Hong Kong at Tenable, said many organisations still prioritise deployment speed over governance.

"AI Appreciation Day shouldn't just celebrate productivity; it should also be a reality check on how we manage risk. The defining feature of this era is the sheer speed of AI adoption. It has created a corporate culture driven by instant gratification, where timelines that used to take days are compressed into seconds. This is unlocking undeniable innovation, but convenience often wins until the consequences catch up.

"Right now, speed has become the overriding priority. Anything that introduces friction, whether governance, compliance, or essential security checks, risks being seen as an obstacle rather than a necessity. The concern is no longer limited to employees experimenting with new tools. AI systems are increasingly being connected to corporate applications, data, and workflows, effectively becoming digital insiders able to act on an organisation's behalf.

"Tenable's Cloud and AI Security Risk Report 2026 found that 18% of organisations had AI services with administrative permissions that were rarely audited. It also found that 52% of non-human identities held critical excessive permissions, compared with 37% of human identities. Giving AI access is not inherently the problem. The risk emerges when organisations grant that access without the visibility, oversight, and controls they would demand for a highly privileged employee.

"The scale of the wider exposure is already becoming clear. Over a recent 30-day period, we detected 457 million AI-related security issues across more than 7,000 organisations, averaging about 62,000 exposures per organisation. These were not simply conventional software vulnerabilities. Many were linked to misconfigurations, unmanaged dependencies, and AI tools operating without sufficient oversight.

"In the past, we saw a similar pattern when firms rushed to deploy cloud technologies. The difference is that AI is being adopted even faster and connected more deeply to the systems businesses depend on. This is not an argument against AI; its benefits are real and irreversible. But we are building an exponential future on a fragile foundation," said Wang.

Some security leaders argue that the rise of so-called frontier AI has already transformed the threat environment. Andrew Kay, Senior Director, Systems Engineering APJ at Illumio, said the shift has forced organisations to confront the limits of prevention.

"AI Appreciation Day isn't just a celebration of innovation - it's a stark reminder of how much cybersecurity has changed in just 12 months. AI has accelerated at a pace some predicted but few were prepared for, giving organisations unprecedented opportunities while also handing cybercriminals faster, smarter, and more scalable ways to attack.

"One thing frontier AI has made impossible to ignore is the reality of cyber risk. It has stripped away any ambiguity about the speed of change and what organisations need to do now. Frontier AI capabilities are no longer exclusive or controllable; they exist, they are spreading rapidly, and they are being operationalised by threat actors. Governments and regulators around the world have been unequivocal in recognising this threat in recent months.

"Organisations that still rely on prevention alone are fighting yesterday's battle. In the age of frontier AI, resilience is measured by how effectively you contain inevitable breaches, not by the illusion that you can stop them. If an attacker gains access, the priority is stopping them from reaching critical systems and sensitive data before they can cause catastrophic damage. This is what I hope to see when this day comes around next year - a meaningful shift toward breach containment," said Kay.

Enterprise technology leaders are also using the day to describe how AI projects are moving from proofs of concept into production environments. Mohammed Rafee Tarafdar, Chief Technology Officer at Infosys, said many organisations are now focused on execution at scale and the role of human judgement.

"On AI Appreciation Day, it's worth reflecting on how the conversation around AI has evolved. For many organisations, the challenge is no longer proving AI's potential but realising sustainable value at scale. As adoption accelerates, the focus is shifting from experimentation to execution, and from isolated use cases to enterprise-wide transformation.

"Making AI work at scale is less about excitement and more about strong foundations and disciplined execution. Success depends not just on technology, but on aligning mindset, operating models, cost-aware architecture, and responsible-by-design governance with rapid innovation. Organisations that get this right will move beyond pilots and unlock long-term value.

"At the same time, AI's greatest potential lies in augmenting human capabilities. The most successful organisations will embrace a Human + AI approach, combining human judgement with intelligent systems to drive better decisions, accelerate innovation, and reimagine work. As AI becomes more embedded, trust, transparency, and accountability will be critical. Ultimately, leaders in the next era of AI will be those who balance rapid innovation with strong foundations, responsible governance, and a commitment to empowering people - unlocking AI's full potential to create lasting value," said Tarafdar.