Meta has released its latest Adversarial Threat Report, offering an updated assessment of how scams, influence operations and AI-driven abuse are evolving across global digital platforms. Published twice a year, the report highlights emerging patterns in criminal activity, with a particular focus on Asia Pacific, where cross-border scam networks are expanding in both scale and sophistication.
A central finding in the latest edition is the takedown of a large criminal scam syndicate operating out of Cambodia. Between January and October 2025, Meta removed more than 6,400 assets linked to the group, which impersonated government officials and law enforcement officers to target victims across Australia, Indonesia, the Philippines, Thailand and Vietnam. According to Meta, the operation illustrates how organised fraud networks are increasingly structured, coordinated and difficult to disrupt due to their regional reach and use of advanced tools.
The report also reinforces that scams and influence campaigns are no longer isolated or opportunistic efforts. Instead, they are becoming sustained operations that adapt quickly to platform enforcement, regulatory pressure and user awareness. Meta said these trends require closer coordination between technology platforms and public agencies to reduce harm and disrupt criminal infrastructure.
Scam networks grow more organised across APAC
Meta’s findings show that scam centres across APAC are evolving beyond simple phishing or impersonation attempts. Criminal groups are now routinely shifting conversations away from mainstream platforms to encrypted messaging apps and alternative payment channels, reducing traceability and increasing the likelihood of successful fraud.
To appear credible to victims, scammers are increasingly using stolen or synthetic identities designed to look local. These may include fake profiles that mimic regional naming conventions, accents or cultural references. The report notes that such tactics are frequently observed in Australia, Singapore and India, where impersonation scams and fraudulent investment schemes remain among the most commonly reported threats.
The industrialisation of scams is also a growing concern. Meta observed the use of template-based scripts, automated messaging tools and scalable workflows that allow scam networks to target large numbers of people with minimal effort. These methods enable criminal groups to test different narratives quickly, discard ineffective approaches and replicate successful scams across borders.
According to Meta, this level of organisation makes scams harder to detect using traditional signals alone, as individual messages or accounts may appear legitimate when assessed in isolation.
AI accelerates both abuse and protection efforts
The report highlights the dual role of artificial intelligence in the current threat landscape. Across APAC, scammers are increasingly using AI tools to enhance the realism and reach of their operations. This includes the creation of deepfake celebrity endorsements, which Meta said are particularly common in Australia and India, as well as multilingual job scams designed to target cross-border jobseekers.
AI is also being used to personalise scam outreach at scale, enabling messages to be tailored based on language, location or perceived interests. These techniques lower the barrier to entry for scammers and allow even smaller groups to operate with the reach of much larger organisations.
At the same time, Meta said it is deploying AI defensively to counter these threats. The company has expanded behavioural detection systems that analyse patterns of interaction rather than relying solely on static content. Users are also receiving more alerts when potentially risky behaviour is detected, helping them identify and avoid scams earlier.
Additional safeguards include tighter insider-threat controls and new protections for Meta’s own AI systems. These include the Llama Firewall and the Rule of Two framework for AI agents, which are designed to reduce misuse and limit the potential for AI-enabled abuse.
Regional and global actions to protect users
Beyond platform enforcement, the report outlines Meta’s continued collaboration with regional authorities to address organised scam activity. In Singapore, Meta supported arrests linked to coordinated scam networks as part of ongoing cooperation efforts aimed at disrupting criminal operations.
To reduce impersonation scams involving public figures, nearly 500,000 individuals have been onboarded globally to Meta’s facial recognition programme. The initiative is intended to limit fraudulent accounts that misuse the likenesses of politicians, business leaders and celebrities, a problem that remains widespread in Australia, Singapore and India.
Globally, Meta said it continues to remove large volumes of fraudulent accounts and work with partners across the financial and public sectors to identify and disrupt scam activity. The company noted that sustained, coordinated action remains essential as criminal networks continue to adapt and exploit new technologies.
