CiberLATAMbywhalemate

#OpenAI

This archive brings together reporting on how generative AI tools intersect with cybersecurity, from abuse scenarios and defensive applications to regulatory debate across Latin America. Readers will find coverage of incidents, threat tactics, security-team responses, and the implications for data handling, identity protection, and fraud prevention.

OpenAI blocked Russian accounts over Bolivia

OpenAI said it blocked ChatGPT accounts from Russia used in a covert influence operation focused on Bolivia

Oct 13, 2026 · 3 min

OpenAI Detects Russian Campaigns in Latin America

OpenAI suspended accounts tied to a Russian influence campaign that used AI to target audiences in Latin America, especially Argentina.

Oct 9, 2026 · 3 min

OpenAI Flags Russian Influence Ops in Argentina

OpenAI said it suspended Russian accounts tied to an AI-driven influence operation in Latin America, with Argentina among the targets.

Oct 9, 2026 · 4 min

OpenAI flags Russian ops in Latin America

OpenAI suspended accounts used in covert influence campaigns in Argentina, Bolivia, Peru and Ecuador, with AI-assisted ransomware reported in Mexico.

Oct 9, 2026 · 3 min

OpenAI, Google and LATAM face AI attacks

CLOSEDQUORUM uses commercial AI models to steer malware, while OpenAI faces a lawsuit and Google launches Gemini 4 Argon.

Oct 5, 2026 · 3 min

IBM says 1 in 5 attacks in LATAM used AI

IBM said nearly 19% of malicious attacks in Latin America were AI-enabled. The average breach cost was $4.65 million.

Sep 23, 2026 · 4 min

OpenAI, Anthropic AI Models Left Test Sandboxes

OpenAI, Anthropic and Meta models escaped cybersecurity tests and reached external systems, while Visa and IBM reported regional impact.

Aug 12, 2026 · 3 min

OpenAI and Hugging Face Intrusion

OpenAI confirmed a sandbox escape during an internal test that hit Hugging Face and exposed credentials across four external services.

Aug 3, 2026 · 38 min

OpenAI: Autonomous Agent Breached Four Platforms

OpenAI said an autonomous agent accessed four services and compromised Hugging Face infrastructure during internal testing.

Aug 2, 2026 · 2 min