Auto-complete is a wonderful feature, when it works correctly.
But when it saves data from fields it shouldn't, or provides suggestions that don't match the field, however, it burdens users with a tool that simplifies life in some areas, but complicates it in others - the complaints about Edge saving and suggesting data at inappropriate times, and drop-downs blocking page content aren't new.
There have been numerous inquiries on Microsoft support platforms, from consumers and commercial users alike, about the ability to restrict auto-complete without fully disabling the feature, or workarounds for when it interferes with user ability to interact with the page content.. The few suggestions made (beyond disabling the functionality entirely) do not resolve the issue.
In our case, auto-complete has created a significant disruption by injecting data where it doesn't belong, invalidating integrity of a simple bulk data entry process that's been unchanged and problem-free for decades in IE. It's been wreaking havoc with business critical processes, turning what used to be a time-saving methodology into a more labor-intensive and error-prone approach than doing things with old-school pen and paper.
I suspect the negative influence of auto-complete is more widespread than community postings might indicate - it took us months to identify auto-complete as the culprit behind the introduction of inaccurate data into our systems, not to mention significant capital invested in troubleshooting, replacing, and updating the data entry equipment that Support insisted was the cause.
Long story short: The implementation of auto-complete in Edge is too aggressive and detrimental to basic functionality and usability of the browser to not to have some method of being able to rein it in; the ability to restrict multiple other browser functions by domain and URL has been standard for decades, making it all the more frustrating that this capability doesn't extend to auto-complete.
Many other businesses and users already disable auto-complete for security reasons, to prevent the caching of specific data fields that contain sensitive information. The quirks introduced in Edge will only have increased that number.
Forcing a binary decision between data integrity and security versus improved efficiency of repetitive data entry is a lose-lose situation for Edge. It either denies a large subset of users access to a basic browser functionality, or introduces significant complications that make using the browser incompatible with users needs.
Developers, it's past time to expand the ability to fine-tune the auto-complete functionality. Basic restrictions to enable/disable based on domain would go a long way in easing the user experience and business operations burden that the current implementation creates. Take it further by introducing functionality that would allow field-specific configuration - custom association of misrecognized fields to a different dataset, for starters, but consider also clearing only certain entries on exit, expiration of saved data by elapsed time since last use, removal of the least-used entries at set intervals, use statistics, string-specific restrictions, export/import capability... there a dozens of small functionalities that could provide significant benefit to end users, especially as more and more applications move to browser-based service models.
Best regards, and much appreciation for taking the time to read this,