Viewed through Bayesian lens, the problem with this article is pretty obvious. You could almost define "speculative" as "giving an observation for which the priors are unclear and only vaguely gestured at".

Here, we have the observation that Sweden gave $60 million foreign aid to Cambodia, but we are given no clear idea of the prior probability that Sweden would give this much and at this time (but we are given the implication that the evidence is significant). We don't need a hard probability estimate (few of us are actually going to rigorously apply Bayes' rule), but we do need enough information to come to a basic idea of the probability on our own. What does Sweden usually give to Cambodia? When is this usually announced? If this is an unusual amount of money, are there other good explanations for this besides corruption?

The problem is that the data in the article leaves us to use a relatively uninformed prior. $60 million is not that much money for a nation with Sweden's economy. It seems fairly likely that some kind of diplomatic news mention would happen regarding Sweden and Cambodia within any given two week period (close enough to appear associated with any given event), and since Sweden and Cambodia are on friendly terms, that mention is likely to be positive. Supposing that this is an unusually large amount of aid, there are plenty of possible reasons that Cambodia, often plagued with both droughts and floods, might have been hit particularly hard in the last year and needed more, so even if this is an unusual amount of aid. This all puts the prior for the observation pretty high, which cuts its evidential weight down very very low. As you illustrated, that prior gets higher as we add information, further weakening the evidence.