Overview
- Akamai forecasts that large language models will become sources of new attacks, citing prompt injection, training-data poisoning and manipulation of external data feeds as emerging techniques.
- The company also predicts a shift to organization-focused fraud, with real-time deepfakes realistic enough to fool seasoned professionals during live calls and virtual meetings, forcing a rethink of identity verification.
- Globant reports that 97% of companies have already suffered AI-related security incidents linked to weak access controls, noting moves toward predictive, AI-driven defenses that still falter without strong governance.
- PwC finds more than half of security leaders are exploring or deploying autonomous AI, with 27% planning implementation within two years, while warning that oversight and ethical controls remain essential.
- Fortinet highlights a shortage of 329,000 cybersecurity specialists in Latin America and a global gap exceeding 4.7 million, with high intrusion rates and training deficits compounding risk as unsanctioned 'shadow AI' use grows when adoption is delayed.