In a recent Twitter thread, she shared information from a former Twitter employee about how the company uses AI to assign "scores" to users based on the words they use.
For example, her twitter "source" says people who use the phrase "2000 Mules" are assigned points by Twitter every time they use it. People who repeatedly used certain words (ballot, fraud, patriot, hacked, etc.) were likely spreading "misinformation." So Twitter taught their AI to assign a score for words/phrases. Accounts with 300 points are reviewed either by AI or by a human moderator.
Don't know who she is, but apparently she's been on Bannon's show and is a television personality that contributes to Fox Business, Good Morning America, Inside Edition, and owns a public relations company in NY.