The cybersecurity landscape across Asia is continually reshaped by rising cyber threats and the rapid adoption of Artificial Intelligence (AI) technologies. Whether private or public, organisations face increasingly sophisticated attacks amid geopolitical tensions and hybrid work models.
Proofpoint CEO Sumit Dhawan underscores the collaborative potential of agentic AI when combined with a human-centric approach and unified data defence. This partnership will set a new standard for enterprise security, enabling faster threat detection, automated data discovery, and adaptive protection tailored to diverse regulatory environments across the APAC region.
Powerful combination
The rising cyber threats in the Asia Pacific (APAC) region have driven the adoption of AI-powered security solutions. However, as new technology helps strengthen a security posture, threat actors also utilise technologies like AI and cryptocurrency for malicious purposes. Dhawan describes the shift as "mind-boggling and fascinating," observing that generative AI and cryptocurrencies have become a powerful combination in recent years.
"All of a sudden, no currency is any longer immune and no language is a barrier," he said.
As a result, Dhawan observed that natural language threats and socially engineered threats have increased by seven to eight times in Japan, and have also multiplied in other countries.
"Those are the two predominant reasons because there's no way all of a sudden threat actors learnt how to speak Japanese or how to take Singaporean dollars sort of easily and be able to use it globally. Those are two trends that are making everyone say that every enterprise across the globe is no longer immune to threat actors," Dhawan said.
Because of this, he believes that enterprises in the region need the most sophisticated protection, beyond the typical basic network-based security, to protect against these sophisticated threats.
Trust in AI-powered security solutions
Asked if he can trust AI as a significant part of an organisation's security strategy, he agreed, believing that agentic AI can help accelerate the automation processes.
"I think the scepticism of whether AI will work in detecting threats versus not still exists," he admitted, but he finds the perception changing gradually.
"I wouldn't say a year later, all of a sudden everyone is convinced that AI can solve it, but our data suggests that it can," he said.
Data defence
In Dhawan's assessment, data security is a fundamental challenge among enterprises, saying that "the problem is the fact that data sits in so many places and is such a vast volume of data."
He also realises the impracticality and the near impossibility for humans to catalogue and classify data: "There are not enough humans that exist in an enterprise to be able to do it, and classifying data is purely a language problem."
Dhawan believes that agentic technology is designed to address these problems. Admitting that expecting agentic technologies to do everything would be futile, the technology can "mix automation with a language model and apply it to perform a lot of repetitive tasks with some degree of intelligence."
At Proofpoint, he shared that they use agentic AI in a supervised fashion, by sampling the data, letting customers look at the sample, assessing the models, tuning the models based on business needs and then letting agents run using that tuning "to do all the work that they can never do themselves".
"It's like a force multiplier of what humans can do and does the work that just humans practically won't be able to do because there's just so much of it. In the absence of agentic AI, the world is what it is today. No enterprise knows where its data is. No enterprise knows how it's classified. No enterprise knows who's accessing it," he said.
Addressing the cybersecurity talent shortage
"The threat landscape and cybersecurity requirements are growing much, much faster than talent can ever match. Not to mention that there is already a shortage that exists, so the practical approach to handling this is more and more of automation," he said.
He believes that agentic AI is the next and best form of automation, with its sense of intelligence, rather than just rule-based automation.
"Agentic AI can handle all the processing of those alerts or reports that humans handle. That is the answer where cybersecurity professionals can spend less time operating products and technologies and more time designing cybersecurity strategies and managing risk. That's the best use of time. Not handling a multitude of alerts and user reports, and triaging them. Let agentic technologies handle those.
Quantum computing
Dhawan is certain that quantum computing will have a profound impact on the cybersecurity landscape, although he believes its viability will still be several years away.
"We are in the early stages of assessing the technology and making sure it's in our labs. So while the landscape may get altered with quantum computing, I think it will only benefit from larger players who can make sure that they can adopt the technology when it becomes more viable from an economic and scale perspective," he said.
In the meantime, he said they are "keeping a very close eye on it, investing in it, learning it, but we don't anticipate that this is going to happen in the next sort of 6 to 12 months."
Transforming enterprise security practices with agentic AI
Dhawan reiterated that agentic AI is going to become a force multiplier for cybersecurity.
"No question. It's powerful. It will be built into products and solutions. You can't just have a workforce. Look at all the documents and data that exist everywhere. Agentic AI is the only solution," he said.
He added that the technology also offers a "degree of relief" in addressing the talent and skills shortage, augmenting the capabilities of cybersecurity professionals.
However, he emphasised that the adoption of agentic AI must be secured. Virtual agents, like humans, can also get fooled, lose data and be maliciously engineered.
"We are not just making sure we provide agentic AI as part of our products, but we are also investing in extending our human-centric platform for virtual humans as well," he said.
Recognising that a combination of humans and virtual humans will comprise the workforce of the future, he underscored the importance of extending a human-centric security platform to manage security, prevent threats, and mitigate data loss.
Investing in cybersecurity strategies
To succeed in implementing cybersecurity strategies with agentic AI, Dhawan advises CISOs to consider three key pillars for security: preventive detection and response, and zero trust as the last line of defence.
He added, "Learn the power of AI. It's helping customers globally. Scepticism is normal, but the data will prove how the power of AI, both in terms of threat defence and data security, can change the game in terms of how it can strengthen the security portfolio."
Lastly, he encourages security leaders to embrace new ways of managing security: "Leverage the power of agentic AI to extend the security sort of personnel that you have. To me, that's the future of cybersecurity, and many CISOs I speak with are already there, but it's still something that many are struggling with."
Setting new standards for enterprise security
Modern cybersecurity threats require modern solutions. As enterprises invest in innovative AI-powered cybersecurity solutions, it is best to synergise them with a human-centric approach and unified data defence.