According to ESET research cited by ForkLog, 79% of organizations surveyed in Singapore encountered at least one AI-related cyber threat in the past year. The poll covered 400 cybersecurity decision-makers across the country.

Most common attack types

The biggest headaches? AI-powered phishing and impersonation, reported by 46% of respondents. Attacks targeting AI tools themselves—including prompt injections—came in second at 41%. Deepfakes and voice cloning rounded out the top three, with 39% saying they'd run into these threats.

AI adoption outpaces security controls

AI is already in play: 97% of those surveyed said they’re using or testing AI tools. But only 49% actually monitor who can access these tools or the outputs they generate. That’s a pretty glaring gap between rolling out AI and keeping it locked down safely.

The findings are based on a survey of 400 cybersecurity professionals in Singapore, reflecting trends over the past year.