A examine finds LLMs from Anthropic, Google, OpenAI, and xAI might help with educational fraud, particularly serving to non-researchers submit fabricated papers to arXiv (Elizabeth Gibney/Nature) March 8, 2026