| Level 0 | Level 1 | Level 2 | Level 3 | Level 4 | Level 5 |
| Behavioral risk | |||||
| Only human risk | Human risk but mitigated by ADAS | Human risk but mitigated by ADAS | Driver and AI risk (within well defined -restricted- ODDs**) | Driver and AI risk (within well defined -extended- ODDs**) | AI risk only |
| Internal risk | |||||
| Only standard components | Standard + automation enablers | Standard + automation enablers | Standard + automation enablers | Standard + automation enablers | Standard + automation enablers |
| Cyber risk* | |||||
| None | Low | Low | Medium | High | High |
| Connectivity risk* | |||||
| None | Low | Low | High | High | High |
| HMI risk* | |||||
| None | Low | Medium | High |
Low/ |
None |
* Relative exposure to risks of the different automation levels, assuming that it is possible to measure and disentangle them
** ODD = Operational Design Domain is the set of all conceivable conditions, scenarios, restrictions that an AV may encounter in a predefined area
Source: Swiss Re Institute