Ahead of the US midterm elections, OpenAI says an operations effort involving hundreds of staff will monitor abuse until vote counting ends. Attempts to disguise deceptive photo requests as cartoon transformations are among the patterns being examined. Anthropic describes analyzing sequences across fake news outlets and multiple accounts rather than judging isolated prompts alone, and says it has blocked associated accounts. The announcements illustrate signals providers use to detect misuse, not independent proof that all misinformation is caught or that electoral effects have been prevented.