OpenAI has introduced BioBasicsBench, a 100-task benchmark built with Ligo, a biotech company focused on enzyme design, to test whether AI agents can handle routine wet-lab analysis. Unlike benchmarks built around figures selected for scientific papers, the tasks use raw evidence from everyday bench work, including agarose and SDS-PAGE gels, Western blots, colony plates, microscopy stacks, sequencing reads, plate-reader exports and videos of lab procedures. The tasks cover five areas: protein methods, imaging and cytometry, quantification and assays, nucleic acid methods, and molecular cloning. Scientists wrote each task from their own experiments and paired it with an expert-written rubric that awards partial credit against individual criteria. Each task passed an internal review, an external review by a PhD-level bench scientist, and a final internal review. Agents must interpret the data, catch errors such as control failures, and recognise when the evidence cannot support a conclusion. At the xhigh reasoning setting, GPT-6 Astra earned an average of 66.7% of the available credit, compared with 43.8% for GPT-5.6 Sol. OpenAI said the results show progress on routine lab analysis but point to remaining gaps in interpreting images, checking controls and applying experimental context. For Australian businesses and professionals, the immediate impact is limited. The benchmark is a research measurement, not a product, pricing change or new regulatory obligation, and it does not trigger any Privacy Act requirements. Biotech and research labs in Australia may find it useful when judging whether AI tools can help with routine analysis. The results also show that even the top-scoring model missed about a third of available credit, so any AI-assisted reading of gels, images or controls still needs review by a qualified scientist before conclusions are drawn.