The other party is being tested for cooperation. Amazon raises the price and then reverts if competitors don't follow suit. They didn't revert after seeing sales drop or something, it sounds like they reverted based on a lack of "cooperation" from competitors.
This is tricky stuff to define wrong doing. What if a company wants to see the going rate for a product and just looks to Amazon to get an idea? You know, because a lot of people will shop at Amazon by default unless there is a reason not to, like saving a bit of money. These kind of algorithms become anti-consumer the more they get automated, but they may seem reasonable on the surface or in isolation.
If two algorithms are fine in isolation, but when used together cause overall market prices to rise what should we think of that? In the above example, Amazon would raise their price and the competitor would follow but not quite to the level of Amazon. Then if markets really are competitive (and fast) someone else may step in at a lower price than either, but I don't believe a lot of markets are fast or efficient when it comes to lowering prices.
Now it's possible that there was some separate agreement, for example Amazon and a competitor may have both agreed to raise a price some percentage and then they made small adjustments to look like it was just an automatic response. But the fact that algorithm seems to have accidentally lowered prices on some occasions would seem to suggest it genuinely was acting unilaterally. Or perhaps those failures were a deliberate part of a larger ruse. Ultimately it will be for a jury to decide but the FTC has its work cut out for them.