Colleges and Companies Grapple with ChatGPT Cheating as Researchers Expose Flaws

ChatGPT generates incorrect answers for over half of software engineering prompts, but humans struggle to identify the errors.
Educators are returning to paper exams and requiring editing histories to prevent AI cheating after a surge in academic integrity violations.
AI detectors are currently unreliable in identifying ChatGPT-generated text, leading to concerns about false accusations.
Hackers are uncovering vulnerabilities in ChatGPT that allow it to provide harmful instructions, prompting fixes from developers.
Some companies are banning ChatGPT over data leakage fears, while others are embracing it to improve workflow efficiency.