When Elon Musk tweeted Friday that his deal to buy Twitter was "on hold" as he looked into the extent of Twitter's bot problem, he was poking an open wound at the social media company.
Twitter's challenges with bots and fake accounts have been around as long as the 16-year-old social media service. In 2016, a Russian troll farm used more than 50,000 bots to try to sway the outcome of the presidential election, and Twitter executives have promised to fix the issue. But even as the company says it has made significant progress in eliminating more fake and spammy accounts than ever, experts say artificial intelligence advances are spinning up new ones that are ever harder to detect.