Make a collection of (royalty) free images representing the idea of AI / AI safety / AI x-risk / AI risk that aren’t anthropomorphizing AI or otherwise misportraying AI (both by searching for existing images and by creating more). This could be used by the media, local AI (safety) groups, etc. I think this is less of a problem than it used to be, but still think this could be valuable. If you want funding for that, you could consider applying for a grant from the Long-Term Future Fund: https://app.effectivealtruism.org/funds/far-future.
More info here: https://docs.google.com/document/d/1tAx1d9ePD05nEhZGZz204xzJ42ErA6UePzunAj1ry38/
Contact email: firstname.lastname@example.org