Gravitational wave astronomy has now entered a mature phase, with second-generation gravitational wave detectors currently performing their third observation run and detecting several signals per month. The instruments' sensitivity is limited by quantum noise in a large fraction of the bandwidth. Already 40 years ago, it was realized that such noise is originated by vacuum fluctuations entering the detector unused port. A challenging strategy proposed to reduce such noise consists in the injection of manipulated states of vacuum, known as squeezed states, where the uncertainty in one of the quadratures (phase or amplitude) is reduced at the expenses of the other one. Recently, after decades of experimental developments, this technique has been implemented in LIGO and Virgo, bringing a major increase in their sensitivities. In this talk I will introduce the fundamentals of quantum noise and squeezed vacuum states and describe their implementation in GW detectors. I will also discuss the future developments of this technique. In particular, I will discuss the use of frequency dependent squeezing to obtain a broadband quantum noise reduction. I will conclude my presentation with the status of a prototype for the production of frequency dependent squeezing, developed at the National Astronomical Observatory of Japan, in the former TAMA300 site.