So USA cases are getting back up towards their March-April peak. They stopped decreasing 30-40 days ago, and started increasing again 20 days ago.
In face of this increase, deaths continue to decrease. Despite increased recognition now of the phenomenon of "unreported deaths" in nursing homes etc. So deaths are becoming disconnected from cases.

Has the virus changed? No.

Have we discovered an effective treatment? No.

So why is the virus becoming less lethal?

Look at the recent USA data is saying-are
1. Steadily increasing numbers tested
2. % tests positive is constant for last 6 weeks.
3. Therefore number of cases rising
4. Fatalities dropping steadily – with no hint yet that fatalities are increasing with increased cases (so called "2nd wave") So fatalities (numerator) dropping as number of cases (denominator) increases.
5. Therefore we are finding more of the mild cases with more testing. These mild cases were present in March and April. They just were not found amongst the chaos and with the very limited testing available then.
6. This has burnt through the tri state area, and numbers are falling. Dramatically. If this had years to run, tri-state infections would still be high.
As we find more cases, the apparent mortality rate drops. The infection mortality rate in 25 studies is way under 1% and closer to 0.5%.
Data from Europe is the same.
The average flu mortality is in the order of 0.08-0.1%. Across a population. But it is almost zero in younger people, and is a few % in the elderly.
A recent met-analysis reviewed 25 studies and came up with a best estimate IFR for COFID of about 0.60% (although there were limitations in this meta-analysis suggesting the true figure is a bit lower).
The CDC “symptomatic CFR” is around 0.4%
But both of these figures are population averages and do not take account of age. Which is like saying that each Australian has (on average) 1 testis and 1 ovary and ignoring the importance of gender. The average is mathematically true, but almost meaningless.
A recent Swedish study looked at case fatality rates and infection fatality rates
The CFR was 26%. This was 432 deaths from 1,667 cases. Sounds pretty bad.
However they estimates that for each diagnosed case, there were an additional 44 undiagnosed cases.
This 44 x increase in the denominator decreased the IFR to 0.6% overall.
Using a single cut-off of 69, the IFR < 69 was 0.1% (this is skewed downwards by all the young and middle aged) and the IFR > 70 was 4.3% (this is skewed up by those in their 80's and 90's).