Oracle's AI integration challenges and progress

Oracle's co-CEO, Clay Magouyrk, revealed during an internal town hall that the company's AI rollout faced significant hurdles last year. Despite spending billions over the past two years helping other companies with AI, Oracle struggled to make generative AI useful for its own workforce. Recently, after deploying ChatGPT and Codex, employees are writing code faster, but new bottlenecks have emerged. Magouyrk noted that the company had not yet figured out how to make AI broadly useful for its workforce a year ago. Oracle's Chief Information Officer, Jae Evans, stated that the rollout of ChatGPT Enterprise and Codex in April and May led to 80% adoption within three months. The company implemented corporate standards, security controls, and internal policies. Evans mentioned that the high cost of OpenAI's GPT-6 Astra, which is 2.5 times more expensive than other models, prompted Oracle to provide visibility into model usage and costs. Anthropic's Mythos Preview model, which Oracle tested, uncovered more security issues in its first two weeks than in an entire year, though it generated a high number of false positives. Evans highlighted that while AI has accelerated software development, it has not yet improved the speed of product delivery to customers. Oracle is still working on redesigning its testing, validation, and deployment processes. The company also received early access to Anthropic's Mythos, which is considered one of the most powerful AI systems, with capabilities surpassing those of other models in cybersecurity.