Suppose, Alice has a Billion of photons in random states, while Bob has their entangled pairs. Suppose, Alice can measure the spin of the photons in X axis, while Bob can measure their spin in an axis which is about 60 degrees rotated from the X axis.
If Bob measures his photon spins before Alice measures hers, there is a 50% chance that his photons will point to the up direction. However, if he performs the same test after Alice measures her photons, due to the fact that Alice's measurement affects his, his chance of detecting number of photons in the up direction reduces to 1/4. Reference: https://en.wikiversity.org/wiki/Bell%27s_theorem/Inequality
Now, if, at certain time, Bob decides to measure his photons, he can deduce whether or not Alice performed a measurement. Thus, one information is transmitted faster than light: Alice made a measurement. What's wrong with this argument?