OpenAI’s o3 benchmark has sparked controversy, drawing comparisons to Theranos due to claims of record-breaking performance on the FrontierMath benchmark while having undisclosed access to much of the test data. Epoch AI’s Tamay Besiroglu admitted to a lack of transparency regarding OpenAI's involvement. Experts, including Gary Marcus, have questioned the legitimacy of OpenAI's results, emphasizing the need for independent evaluation. Despite skepticism, OpenAI's CEO remains optimistic about future releases.