(Reuters) - A proposed class action accusing Microsoft's ( MSFT ) LinkedIn of violating the privacy of millions of Premium customers by disclosing their private messages to train generative artificial intelligence models has been dismissed.

The plaintiff Alessandro De La Torre on Thursday filed a notice of dismissal without prejudice in the San Jose, California federal court, nine days after suing LinkedIn, and after the company said the lawsuit had no merit.

De La Torre accused the business-focused social media platform of breaking a promise to use personal customer data only to improve its services, by sharing customers' messages with third parties involved in AI.

The complaint said LinkedIn revealed the unauthorized sharing when it updated its privacy policy in September, and said a new account setting to prevent data sharing would not affect previous AI training.

"LinkedIn's belated disclosures here left consumers rightly concerned and confused about what was being used to train AI," Eli Wade-Scott, managing partner at Edelson PC, which represented De La Torre, said in an email on Friday.

"Users can take comfort, at least, that LinkedIn has shown us evidence that it did not use their private messages to do that," he added. "We appreciate the professionalism of LinkedIn's team."

In a LinkedIn post on Thursday, Sarah Wight, a lawyer and vice president for the company, confirmed that LinkedIn did not disclose customers' private messages for AI training. "We never did that," she said.