After about a week of fairly significant drop in new cases in Oklahoma, it was reported today that yesterday we hit 119 new cases. Not the highest since the start, but the highest in more than a week...right about the time people started letting down their guard and having parties and cookouts for Memorial Day.
Also, it appears to me when watching coverage of the protests, a good many of the individuals are wearing masks.
Reporting of new cases can be kind of sporadic based on how counties report, how busy labs are, etc, etc. Some days a lab might process way more test kits than another because some maintenance was needed on some equipment, limited hours on a weekend, or whathaveyou. In the case of New Mexico, there was a huge spike in case numbers yesterday because they tested a prison and reported all of them at once. The day to day case reports are full of noise.
All of that is to say that it is probably better to look at something like the 7-day moving average. The New York Times has this plotted for all fifty states, with a nice breakdown of where that number is increasing, decreasing, etc. They list Oklahoma as a state where it is decreasing. You can quibble with their metrics if you want, as they have some states as "increasing" by virtue of very low caseloads, like Idaho. Increasing from 1.5 cases per day average to 2.0 cases per day average is a statistically significant rise I guess, but most people would exclude results like that due to small sample sizes. To their credit, they also show the graph and you can click to see the raw numbers, so you can draw your own conclusions from any of their charts. Anyhow, here is the link:
https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html#states
Another thing you can look at that is reasonably indicative of how things are moving is the hospitalization numbers.
Edit:
Also, Georgia is hard to interpret, because while 49 states count a 'new case' on the date that the test come back positive, Georgia backdates the "new case" to the date that the person began exhibiting symptoms. I think asymptomatic people go on the date of the test administration. I think there is probably some merit to doing it this way as it will likely give you a more accurate picture of the overall trendline, but only for the older data. That is, it is no good for real-time analysis on the trendline for the most recent week or so.