And perhaps most troubling of all, the feature they had exploited was introduced into Intel chips in the mid-1990s. The attack had somehow remained possible, without any apparent public discovery, for decades.