RANDOM.ORG generates true randomness via atmospheric noise.
This page shows the source purity statistics.

(Each graph is from a different radio. Click on the graphs to enlarge them.)

The graphs on this page gives an indication of how good the raw
random data used by RANDOM.ORG is. Any source of raw randomness may
be biased (skewed) towards 0 or 1, meaning that more 0s occur in the
data than 1s or vice versa. An important part of generating true
random numbers is to perform skew correction on the raw
random data, i.e., remove any such skew. The graphs on this page
show how close to unskewed the raw random data used by RANDOM.ORG is. If this data is perfectly unskewed, even before process, you
will see a purity rating close to 100%. If there is skew in the
data, you will see a lower rating. However, even if the raw data is
skewed, the skew correction algorithm will correct it, and the skew
will not appear in the processed data.

Each graph shows how a given radio performed on
a particular day. New graphs are generated automatically shortly
after midnight (UTC) every day. Each radio has its own name (e.g.,
copenhagen-hw0), and each graph is labelled with the name of the
radio to which it belongs. Not all radios are active on all
days.

RANDOM.ORG uses a skew correction algorithm devised by John von
Neumann, which considers the bits from the raw source in pairs.
If the two bits in the pair are identical (i.e., 00 or 11), both
bits are discarded. If the two bits are different (i.e., 01 or 10),
the first bit is added to the stream of processed bits. This is a
very simple algorithm, but it isn't very efficient because it
results in the loss of three quarters of the raw random data, even
if the source is already perfectly unskewed.