OpenAI announced that one of its newest A.I. technologies solved a million-dollar math problem, sparking controversy. New York University professor Tristan Buckmaster wondered if OpenAI used his research to help solve it. OpenAI initially acknowledged it "cannot rule out" that data from his use of its products improved its models, but later stated it was "impossible" for Dr. Buckmaster's prompts to have influenced the system.

Dr. Buckmaster's fears highlight a broader issue affecting anyone using A.I. technologies operated by OpenAI, Anthropic, Google and Meta. As people input private information into these systems, they may be feeding future technologies. Oren Etzioni, a professor at the University of Washington, warned that you cannot assume an idea simply disappears into vast data without impact.

OpenAI allows users to turn off a setting that feeds data into new technologies. Aneesh Muppidi, an A.I. researcher at Stanford University, noted that while data can be used to train new systems, not all data gets included. Businesses and research organizations often negotiate contracts preventing OpenAI from using their private data for training.