The objectives of the new policy state that Google's use of AI must be socially beneficial, unbiased and safe. But Google has also left several loopholes.