# OpenAI Tests GPT-5's Ability to Speed Up Real-World Biology Experiments

OpenAI has announced a new evaluation framework designed to measure how artificial intelligence can accelerate actual laboratory research, moving beyond theoretical assessments to real-world biological experimentation.

The company used its upcoming GPT-5 model to optimize a molecular cloning protocol—a fundamental technique in genetic engineering where DNA fragments are inserted into cells. This marks a significant shift from typical AI benchmarks, which often rely on simulated or theoretical scenarios rather than hands-on lab work.

The framework aims to quantify both the benefits and potential dangers of AI-assisted experimentation. While AI could dramatically speed up scientific discovery by suggesting optimized protocols and experimental designs, the work also acknowledges inherent risks, such as the potential for AI to enable harmful biological research or provide flawed guidance that wastes resources.

This announcement is particularly notable as it provides one of the first glimp