here is what my ai chatbot lumo says:
Yes, there is clear evidence of this on the internet. The problem is well documented:
## The Facts
**The actual bug:** On Launchpad (the HPLIP bug tracker), the issue was reported as bug [#2172090](https://launchpad.net/hplip): *"hpaio: chunked http_read() double-counts bytes_read → LEDM scan over USB fails."* Specifically, the `http_read()` function in `scan/sane/http.c` counts read bytes twice, which causes the error **"response exceeds buffer"** during scanning (e.g. `read=8174 len=28 max=8192`). Devices communicating via the LEDM protocol are particularly affected — typically HP DeskJet and OfficeJet All-in-One printers, mostly when connected via USB.
**Confirmations from the forums:**
- On the [openSUSE forum](https://forums.opensuse.org/t/hp-scanner-no-longer-working/196317), several users report that after updating to HPLIP 3.26.6, the scanner is still detected by `scanimage -L`, but every scan attempt fails with **"Error during device I/O."** A downgrade to **3.26.4** fixed the problem for them. There is also a corresponding thread on the Arch Linux forum.
- The bug occurs on various devices — the Launchpad bug tracker includes reports for, among others, the HP Envy 5032 (*"scanning does not work since 3.26.6"*, bug #2169330), OfficeJet Pro 8710/8720, and the Smart Tank 500 — so classic AIO models like the DeskJets.
## What you can do
1. **Downgrade to HPLIP 3.26.4** — the solution confirmed by several users, provided your distribution still offers that package.
2. **Alternative: use ipp-sane/AirScan** — the SANE eSCL driver (`sane-airscan`) completely bypasses the broken HPLIP backend. One openSUSE user solved their problem this way without a downgrade; for this you simply need to set up the scanner via AirScan instead of `hpaio` (e.g. an entry in `/etc/sane.d/airscan.conf`, or just disable `hpaio` in `dll.conf` if AirScan finds the device automatically).
3. **Wait it out** — the bug has been reported on Launchpad; presumably it will be fixed in a subsequent release. There are already patches under discussion.
In short: yes, the update to 3.26.6 introduced a genuine regression bug in the scanning backend that causes exactly the symptoms you described. Printing is not affected — only scanning through the `hpaio` backend.
Do you need a concrete guide for the downgrade on your system (FreeBSD?) or for switching to AirScan?