MANILA - Internet users need to adopt a “zero trust mentality” when it comes to dubious messages and offers online as deepfakes go mainstream in the coming year, experts from global cybersecurity firm Palo Alto Networks said during a media briefing.

Palo Alto said the Asia Pacific region “will face a perfect storm of AI-driven cyber threats, escalating in scale, sophistication, and impact” in 2025.

Steven Scheurman, regional vice president of Palo Alto Networks for ASEAN said the company sees several trends in cybersecurity for the region in 2025, among which is that cybercriminals will use “ever-improving generative AI” to launch credible deepfake attacks. Scheurman said the use of audio deepfakes will also become more widespread in these attacks, as the available technology already has credible voice cloning.

To protect themselves, internet users and businesses need more than just the best cybersecurity tools and technologies–they also need to distrust what they read, hear or even see online Scheurman said.

“So you can have the best tools, the best technology, you know, leveraging AI and so forth. But at the end of the day, there is a human element here, in terms of always having a mindset, which we took, called zero trust,” Scheurman said.

Scammers may target a company by phishing an employee with a call supposedly from a ranking official and get the employee to divulge sensitive information, access, or even money. Employees must be taught to distrust dubious calls, he said.

“Is it normal for my boss to call me, to immediately transfer $350,000? To urgently call a deal? Has he ever done this before? Is this normal behavior? Why is he suddenly doing this now? Having this mindset is critical because this is actually what really happened in the number of cases that I've been involved in,” Scheurman said.

“So the mindset and the mentality–this is something that's not technology-related,” he added.

This kind of cautiousness needs to be cultivated not just among employees of a company, but with the public in general, especially vulnerable groups like the elderly and the young.

“We are now in the festive period in in December, end of the year, and I'm sure in the Philippines there are lots of special offers and sales and promotions and festivities going on. And many of you will receive emails and messages promising you all sorts of special gifts and etc. Do not believe everything you get, particularly if the news is not so overwhelmingly positive, there's got to be a scam or something trying to take advantage of you,” Scheurman added. 

DEEPFAKE PROPAGANDA

Cautiousness should also extend to political messages on social media according to Jobert David, system engineering manager at Palo Alto Networks.

“So basically, I think AI-driven propaganda is now a growing concern in terms of these elections,” David said. A stronger regulatory framework is needed especially on social media. The government should also come up with media literacy programs to inform people about fake news and deepfakes.

“I think what's also important is to have that public awareness campaign that you know, as I said, not everything you see is believed, especially in social media,” David said.

Businesses should also be aware that ransomware is getting built at a much faster pace thanks to AI, according to  Oscar Visaya, Country Manager of Palo Alto Networks in the Philippines. He noted that two years ago, it could take 12 hours to build ransomware, but today this is down to just 3 hours. By 2026, ransomware could be built in as short as 15 minutes because of AI.

“Humans cannot win against machines, against AI, so it won't work. So it has to be machine versus machine, and it has to be AI versus AI and with AI,” Visaya said.

Scheurman said this was the reason why Palo Alto Networks forecasts the shift to unified data security platforms, emphasizing the need for integration and proactive threat prevention.

The company also forecasts quantum security, with bad actors collecting data for future decryption using quantum computing.

“Bad actors are starting to collect a lot of data, and they do this regularly. They're collecting the data with the belief in the premise, as quantum computing becomes more advanced and mature, they'll be able to use quantum computing, basically, to decrypt all of this data that currently, at the moment, can't be compromised,” Scheurman said.

Palo Alto also sees a growing demand for more transparency in AI, along with the need for understanding the logic and ethics behind AI models.