Insurance is a better incentive structure for alignment R&D than philanthropy. Mechanically, insurers have skin in the game because they pay when stuff blows up. This makes them both keenly interested in R&D to prevent stuff from blowing up and in not getting captured by any particular company. This incentive counteracts the financial capture donations can create. Insurers can sub-contract R&D work to existing organizations doing excellent technical work. (Props to @METR_Evals for getting funded without severe conflicts). Historically, insurers have funded lots of safety R&D. For example, car crash tests were invented by Insurance Institute of Highway Safety, many thousands of lives saved by fire safety improvements from Underwriters Laboratory, funded by insurers. A bunch of nuclear inspection R&D is funded by the nuclear insurance mutual. Practically, this requires the existence of an insurance structure that covers meaningful risks. Liability concerns are now top of mind at labs; the grounds are now fertile. You don't need traditional insurance capital to be interested to get started - nuclear power plants cover most losses from their pooled capital in their mutual.