Scientists worry AI is stealing their ideas

Scientists are growing increasingly concerned that artificial intelligence companies may be using their ideas without proper acknowledgment. Researchers are turning to AI as a sounding board and tool for literature review, literature review, and even generating entire proofs for their research. However, there are accusations that companies like OpenAI and Anthropic are presenting users' ideas as those of their models.

One mathematician at New York University claimed that OpenAI had stolen his progress, which he had shared with one of their chatbots. Similarly, a biologist at the University of Copenhagen reported a similar situation with Anthropic. Despite these claims, both companies have stated that they did not train their models on the mathematician's and biologist's uploaded data.

However, with the growing use of AI models in scientific research, there are fears that researchers may feel their ideas are being misused without their consent.

Some scientists are now viewing these AI models as competitors rather than tools, leading to a sense of betrayal. Clément Canonne, a theoretical computer scientist, refuses to use these models. Jennifer King, a fellow at the Stanford University Institute for Human-Centered Artificial Intelligence, warns that users on personal accounts may inadvertently provide their information to train the models, as the companies are not transparent about their practices.

While there is no evidence of rogue events causing data breaches in the case of scientific discoveries, the lack of transparency creates a sense of unease among researchers. They face a choice: use the models and risk leaking their discoveries or not use them and risk someone else publishing their work first. This has created an uneasy landscape where many scientists feel they must sacrifice their privacy to remain competitive.