AI in disinformation is no longer a "scare story". It is an operational risk.
On August 25, OpenAI reported that it had banned a cluster of ChatGPT accounts that most likely operated from Russia. The accounts were used to write social media posts and comments promoting an "expert community" called International Burke Institute. The campaign's reach was small, but it was put together with unusual care. For me the geopolitics is not the important part here. The practical takeaway is: as soon as you have content generation in your product or inside your company, you automatically become part of the "content, trust, decisions" chain.
What to put into practice (even in small projects):
- don't ship generation "to prod" without moderation/quality control
- separate the modes: draft for a human != auto-publishing
- log the sources of prompts/content and set limits
- define your "red zones" in advance (finance, legal documents, public statements).
If you are building a RAG/bot/content generator, security and usage policy have to be part of the architecture, not a "we'll add it later" item.