Colombia Sees Rise in AI Phishing

Offensive and defensive AI on the same board

In Colombia, hyperrealistic phishing and polymorphic malware were the most detected cyberattack methods in the first half of 2025, according to Bloomberg Línea. The report placed that finding in a regional setting where generative AI was amplifying attacks across Latin America, with an impact on several sectors, including financial services.

The same coverage said AI was also being used on the defensive side. According to the report cited by Bloomberg Línea, AI engines were helping identify incidents, speed up response, and reduce false positives across the region. The technology therefore appeared on both sides of the conflict, supporting more convincing attacks and faster detection systems.

Deepfakes and synthetic identities

Another report, this time from Infosertecla based on a Sumsub study, said that in Colombia account takeover attempts rose 188% and deepfake use increased 77% during the period analyzed in that study. That point should be read cautiously, since the journalistic reference itself presented it as information attributed to Sumsub's research and not as independently confirmed data in this note.

Even so, the coverage described deepfakes and AI-generated identities as growing fraud vectors in the region, with an impact in Colombia as well within the secondary report. The focus is no longer just on malicious emails or links, but on techniques designed to impersonate identities and build trust with synthetic material.

Technical context from Microsoft

Microsoft also published a post about threats in the age of AI that mentioned social engineering backed by LLMs to generate phishing emails as an example of offensive artificial intelligence capabilities. That material was not focused on Colombia, but it helps explain the kind of automation threat actors are already using to make their campaigns more convincing.

Taken together, the reports cited show Colombia already sits within a regional map where AI is speeding up both deception and defensive response, while deepfakes, synthetic identities, and hyperrealistic phishing continue to stand out as vectors to watch closely.