So this is not different than people who share Google drive docs with company data with public links. It's not a fault of OpenAI or any other company. With enough time even your laptop can do it. How do they 'know' OpenAI breached? Maybe someone had an agent running asking to do a scan on any public docs?
