We’re working under the assumption that “memory safe languages” can’t be unsafe themselves, which isn’t the case.