Let's study your code first. The problem you mentioned is. The following lines:

$arrayKey = array_search($date, array_column($array, 'date'));

...returns false because the $array entry with the date has not been created while processing the .txt file. (The logic for creating array members using array_push is below the code.)

Simple fix to continue moving to the relevant part of the if/else logic has not been defined yet:

if($arrayKey !== false && $array[$arrayKey]){
...

That is, if $arrayKey is not false, continue adding the value to the existing array member. Otherwise, create an array member.

Additionally, there is an issue when processing images, which generates a warning the first time it occurs:

$amount = intval($array[$arrayKey]['collection-size']);

This will fail with undefined array key 'collection-size' because the collection-size key does not exist yet. Fix e.g. using null coalescing operator to set "default zero" before trying to operate on array keys:

$array[$arrayKey]['collection-size'] ??= 0;

These comments fix the bug, but it would be better to separate the "entry creation" in the first instance of the txt or jpg - with an empty array member with the expected key, before performing any txt/jpg specific logic. I would simply use $date itself as the grouping so you can get rid of array_search too. For example, after extracting the date, use:

$array[$date] ??= [
    'date' => $date,
    'caption' => '',
    'collection-size' => 0,
    'has-video' => false,
];

Then modify the rest of the code to match. Your code should not depend on the order in which files are read. The order is not guaranteed. Otherwise, you can always read the list of files into a regular array first, then sort them, and iterate again when applying specific logic.

Simplified code

The actual amount of code required is much less than what you have. Here I have trimmed it for you. I don't have your file, so here's some dummy data:

$files = 
 You can also put the glob files into an array (= list of file paths): 
$file_directory = "images/instagram";
$files = glob(dirname(realpath(__DIR__)).'/'.$file_directory.'/*');
Then iterate as follows: 
foreach($files as $filepath) {
    $filename = basename($filepath);
    $date = strstr($filename, '_UTC', true);
    
    $array[$date] ??= [
        'date' => $date,
        'caption' => '',
        'collection-size' => 0,
        'has-video' => false,
    ];
    
    $ext = strtolower(pathinfo($file, PATHINFO_EXTENSION));
    
    if($ext === 'jpg'){
        // Each JPG increments collection size:
        $array[$date]['collection-size']++;
    }
    
    elseif ($ext === "txt"){
        // We use a dummy here:
        $caption = '---';
        // $caption = file_get_contents($filepath);

        $array[$date]['caption'] = $caption;
    }
}
 Notice how much it shrinks. what happened? 

• We use $date as the grouping index of the array. No more array_search!
• We initialize a default entry for each date. No further inspections or conditions required!
• We ignore "collection size" such as _3 in the file name: just add 1 for each JPG.
• We use nglob and file_get_contents instead of readdir and fopen.
• The order of the files is not important. (Feel free to test and shuffle($files)!)

result:

array(3) {
    ["2022-12-26_14-14-01"] · array(4) {
        ["date"] · string(19) "2022-12-26_14-14-01"
        ["caption"] · string(3) "---"
        ["collection-size"] · int(3)
        ["has-video"] · bool(false)
    }
    ["2022-12-27_14-14-01"] · array(4) {
        ["date"] · string(19) "2022-12-27_14-14-01"
        ["caption"] · string(3) "---"
        ["collection-size"] · int(2)
        ["has-video"] · bool(false)
    }
    ["2022-12-31_18-13-43"] · array(4) {
        ["date"] · string(19) "2022-12-31_18-13-43"
        ["caption"] · string(3) "---"
        ["collection-size"] · int(1)
        ["has-video"] · bool(false)
    }
}