Honestly some straps inside the already pressurized capsule + some CO2 filtering and Crew Dragon 1 is probably much less risky than most of the early human space flights.
We forget how many astronauts and cosmonauts died in the early days (especially if you include pilots of high altitude test flights which tested many concepts later used in the space programs).
There were many additional close calls on either side, averted only by skilled test pilots, ground controllers, and luck.
Cargo Dragon 1 didn't have a Launch Escape System but was otherwise pretty safe.
CRS-7 is an exception, and even then if the flight software had been written to deploy parachutes, probably survivable.
CRS-2 had issues with thrusters but not life-threatening ones.
Gemini and the Shuttle programs also didn't have LES - Gemini had simple ejector seats and the Shuttle had...nothing after the first test missions removed the ejector seats.
I'd argue Crew Dragon was safer than the first few Shuttle launches as well - the first flights had a 1 in 9 chance of failure. Landing the Shuttle wasn't difficult but had to be successful - other options were trying to ditch in the ocean or using ejection seats on the earliest flights. Some abort options were almost certainly hopeless. STS-1 had massive issues which could have lead to a safety problem. Losing only 2 flights in the STS program is somewhat of a miracle. Falcon 9 now has a higher reliability estimate than STS - 0.99 vs. 0.97 [0]
Earlier on in the Falcon 9 program the point estimate was lower - but so too would have been STS's, especially right after Challenger.
[0] http://spacelaunchreport.com/log2020.html
See ACTIVE LAUNCH VEHICLE RELIABILITY STATISTICS and RETIRED LAUNCH VEHICLE RELIABILITY STATISTICS