How to remove the square surrounding legend keys in a plot made by plot_model (ggplot) in R

I'm currently working on creating some complex visualizations using the plot_model function from the sjPlot package in R. Specifically, I'm dealing with interaction plots of linear models, and I'm facing an issue related to customizing the legend. When I plot it, the keys of the legend appear inside squares, and I would like to get rid of those squares. I've read through the documentation, tried various approaches, and searched through existing questions on Stack Overflow, but I'm still struggling to achieve that.

I've attempted several approaches to remove the squares and borders from the legend in my plot. These include using the theme(legend.key = element_blank()) and theme(legend.background = element_rect(color = NA)) functions to make the legend keys transparent and remove the border lines, but unfortunately, these solutions didn't yield the desired result. They either removed the legend keys entirely or retained a gray background with black keys.

I've also experimented with guides(color = guide_legend(override.aes = list(fill = NULL, color = NULL))) and guides(color = guide_legend(override.aes = list(fill = NULL, line = 0))) to control the legend aesthetics, but these approaches did not fix it either.

Some of the posts that I have checked are: https://stackoverflow.com/a/11270029/18255654 and https://stackoverflow.com/a/2249622/18255654

This is the figure that I am currently obtaining:

To test the script and post it here, I am using two available datasets, and I am using the following script:

rm(list=ls())

library(nlme)
library(lme4)
library(ggplot2)
library(sjPlot)
library(ggthemes) # Library of themes for ggplot

# Load dataset for:
data=sleepstudy # Example 1
data(efc) # Example 2

# Example 1
group_colors <- c(old = "#3b71cd", young = "#c0393d")
threshold_subject_number <- 9
data$Subject <- as.numeric(data$Subject)
data$Group <- ifelse(data$Subject <= threshold_subject_number, "young", "old")

analysis_model = lme(Reaction ~ Days * Group,
                     data=data,
                     random=~1+Days|Subject, 
                     control = lmeControl(opt = "optim"), 
                     method = "REML")

line_plot = 
  
  plot_model(analysis_model, # Model to plot
             type = "int", # Type of plot: int = interaction
             title = "Title", # Add title
             colors = group_colors, # Define the colors
             axis.title = c("Day", "Predicted Value")) +
  theme_hc() +
  theme(plot.title = element_text(hjust = 0.5)) + # Center plot title
  theme(legend.position = "right") + # Move legend to right side
  guides(color = guide_legend(override.aes = list(fill = NA, # Make square transparent
                                                  size = 2), # Change size of square
                              reverse = TRUE)) # Reverse order of legend

# Example 2
m = glm(tot_sc_e ~ c161sex + c172code * neg_c_7,
        data = efc,
        family = poisson())

mp = plot_model(m, type = "int") +
     theme_hc() + # Change theme
     theme(plot.title = element_text(hjust = 0.5)) + # Center plot title
     theme(legend.position = "right") + # Move legend to right side
     guides(color = guide_legend(override.aes = list(fill = NA, # Make square transparent
                                                     size = 2), # Change size of square
                                 reverse = TRUE)) # Reverse order of legend

## Different things I've tried:

# From: https://stackoverflow.com/a/11270029/18255654
mp +
  theme(legend.key = element_blank()) # Does not work - nothing happens

# From: https://stackoverflow.com/a/2249622/18255654
mp +
  theme(legend.background = element_rect(color = NA)) # Does not work - nothing happens

mp +
  theme(legend.key = element_rect(colour = NA, fill = NA)) # Does not work - nothing happens

mp + 
  guides(color = guide_legend(override.aes = list(fill = NA, color = NA))) # Does not work - removes also the keys

mp + 
  guides(color = guide_legend(override.aes = list(fill = NULL, color = NULL))) # Does not work - removes line but turns all keys to black and restores grey background

mp + 
  guides(color = guide_legend(override.aes = list(fill = NULL, line = 0))) # Does not work - removes line but turns all keys to black and restores grey background

The outlines are caused by the geom_ribbon used under the hood to draw the confidence band. One option to get rid of the outlines would be to manipulate the ggplot object returned by plot_model, i.e. by setting show.legend=FALSE for this layer aka the second layer using mp$layers[[2]]$show.legend <- FALSE:

library(ggplot2)
library(sjPlot)
library(ggthemes) # Library of themes for ggplot


mp <- plot_model(m, type = "int") +
  theme_hc() + # Change theme
  theme(plot.title = element_text(hjust = 0.5)) + # Center plot title
  theme(legend.position = "right") + # Move legend to right side
  guides(color = guide_legend(
    override.aes = list(
      fill = NA, # Make square transparent
      size = 2
    ), # Change size of square
    reverse = TRUE
  )) # Reverse order of legend

mp$layers[[2]]$show.legend <- FALSE

mp