Visa Launches AI-Powered 'Intelligent Authorization'

Visa's Intelligent Authorization is a new feature on its Visa Acceptance Platform, designed to modernize payment processing for acquirers through a single API connection. Its machine-learning engine analyzes transaction data in real-time to optimize routing decisions. This system boasts 99.999% uptime and helps achieve an average global approval rate of 96.3%. The broader VisaNet +AI suite, which includes Intelligent Authorization, leverages deep learning to enhance payment experiences. One of its capabilities, Smarter Stand-In Processing (STIP), uses AI to make approval decisions on behalf of issuers during outages with up to 95% accuracy. Visa's AI-powered risk models process over 300 billion transactions annually, blocking an estimated $40 billion in fraudulent transactions each year. Cape Town-based Orca Fraud is tackling payment security in emerging markets, where legacy fraud detection systems often fail due to fragmented and unstructured data. Co-founders Thalia Pillay and Carla Wilby designed the platform to analyze messy, real-world data from mobile wallets, cards, and bank transfers across Africa. Orca's machine learning models are trained on these complex local datasets to provide real-time fraud intelligence. Orca Fraud now monitors over $5 billion in monthly transaction volume across more than 70 countries. The company's recent $2.35 million seed round, led by Norrsken22, will fund expansion and enhance its transaction monitoring capabilities in these high-growth, high-velocity payment environments. The push for AI in fraud detection is a major industry trend, with the global market for AI in fraud detection expected to reach nearly $120 billion by 2034, up from $15.6 billion in 2025. This growth is driven by the need for more adaptive models to combat increasingly sophisticated fraud tactics like deepfakes, which have seen a surge of over 1,300% since 2023. AI-driven systems can reduce fraud losses by as much as 50% compared to traditional rule-based methods. For students aspiring to enter this field, building a portfolio with relevant projects is key. Ideas include creating fraud detection models using binary classification algorithms like Logistic Regression or Random Forest on historical transaction data. Another practical project is developing a stock price prediction model using time-series analysis and sentiment analysis of financial news headlines. Los Angeles is an emerging hub for fintech, with companies like Block (Cash App, Square) and Upstart actively hiring for machine learning roles focused on risk and fraud detection. Roles for machine learning engineers in the LA area, particularly those with a focus on fraud, can command salaries ranging from $245,000 to $345,000 a year. Networking with these companies can provide a direct pathway from graduation to a high-demand role in the local tech scene.