In the list of new and resolved bugs in iOS 14, the following is currently written:
"The performance of Apple's new VoiceOver recognition features is inconsistent and unreliable, particularly that of Screen Recognition. It should be noted that these features are a work-in-progress, and your mileage may vary. As things are now, you should expect to find that there will be situations where having one or more of these features enabled will provide a worse experience than having them disabled."
However, the post does not state that on iPhones older than the XS (e.g. X, 8 and 8 Plus, 7 and 7 Plus, 6s and 6s Plus, as well as the 1st generation SE) as well as possible iPods (not really sure about that line because I don't have one), there is no way of disabling this feature.
Contrary to common belief, and I believe this is why this was missed in the beta cycles, older iPhones do not have a VoiceOver recognition menu in it's settings. This is only present in models that properly support this feature. However, since the dawn of iOS 11 I believe, there have been efforts made to recognise images, all be it with a gesture. Then in iOS 13, Apple took it a step further and gave us the ability to know what the possible text of controls was. This setting had a verbosity option, which could be turned off, or a sound could be played to indicate these possible text and images. This option has disappeared, possibly in favour of the recognition options.
So, what's happening on older phones now, since this has changed in iOS 14? Well, whenever VoiceOver detects an image, or a control with an image, there is a sound played, while some text is reported which tries to indicate what's in the image. The text that is spoken is based on the older recognition abilities, so examples would include: adult, necktie, shoes, food, knife, fork, wood processed, and other one-word descriptions. Again, this cannot be disabled. The sound, however, can be turned off in the VoiceOver audio settings, just look for described text and images. This turns off the sound, but not the behaviour.
The implications of this are more harmful than good. It causes text that is not on screen to be spoken. It also leads to extra verbosity, which can slow down productivity. The descriptions may be inaccurate, and there is no warning or disclaimer that this may occur. Apps where this happens include Apple Music, Apple Podcasts, WhatsApp, Facebook, Twitter, Instagram and Messenger, among many others. Many blind persons have iPhones listed above, and would be forced to upgrade in order to stop this, unless Apple fixes this behaviour. What about those that cannot upgrade, though? What about others that do not want to bear the financial burden? I have written to Apple Accessibility about this, and I encourage others with this issue to do the same.