diff --git a/NEWS.md b/NEWS.md
index 006a338..7c1a203 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -13,6 +13,9 @@
        and if so, and installs cuDNN, updates "~/.profile", and emits additional 
        instructions for how to install the necessary CUDA drivers to enable GPU usage. 
        Set new arg `configure_cuda_vars=FALSE` to disable.
+     - `pip_ignore_installed` default is now `FALSE` again.
+     - On Arm Macs (M1/M2), the default tensorflow package is once again installed, 
+       rather than `tensorflow-macos` and `tensorflow-metal`.
 
 - New `pillar:type_sum()` method for Tensors, giving a
   more informative printout of Tensors in R tracebacks and tibbles.
diff --git a/R/install.R b/R/install.R
index 4422f46..0a8c767 100644
--- a/R/install.R
+++ b/R/install.R
@@ -25,16 +25,12 @@
 #'   Sys.setenv("RETICULATE_PYTHON" = "~/path/to/python-env/bin/python")
 #'   ```
 #'
-#' @section Apple Silicon: Tensorflow on Apple Silicon is not officially
-#'   supported by the Tensorflow maintainers. However Apple has published a
-#'   custom version of Tensorflow compatible with Arm Macs.
-#'   `install_tensorflow()` will install the special packages `tensorflow-macos`
-#'   and `tensorflow-metal` on Arm Macs. See
+#' @section Apple Silicon: Beginning with Tensorflow version 2.13, the default
+#'   tensorflow package now works on Apple Silicon. See
 #'   \url{https://developer.apple.com/metal/tensorflow-plugin/} for instructions
-#'   on how to do the equivalent manually. Please note that this is an
-#'   experimental build of both Python and Tensorflow, with known issues. In
-#'   particular, certain operations will cause errors, but can often be remedied
-#'   by pinning them to the CPU. For example:
+#'   on how to install older versions of Tensorflow on macOS. Please note that
+#'   not all operations are supported on Arm Mac GPUs. You can work around the
+#'   missing operations by pinning operations to CPU. For example:
 #'
 #'   ```` R
 #'   x <- array(runif(64*64), c(1, 64, 64))
@@ -232,7 +228,6 @@ function(method = c("auto", "virtualenv", "conda"),
     msg <- "# Configured by the R function tensorflow::install_tensorflow()"
 
     if(!all(c(msg, vars) %in% profile)) {
-      browser()
       profile <- c(profile, "", msg, vars, "")
       writeLines(profile, "~/.profile")
       message("- Updated file '~/.profile' with new environment variables.")
diff --git a/man/install_tensorflow.Rd b/man/install_tensorflow.Rd
index 0848a95..115db65 100644
--- a/man/install_tensorflow.Rd
+++ b/man/install_tensorflow.Rd
@@ -14,7 +14,7 @@ install_tensorflow(
   conda_python_version = NULL,
   ...,
   configure_cuda_vars = NULL,
-  pip_ignore_installed = TRUE,
+  pip_ignore_installed = FALSE,
   new_env = identical(envname, "r-tensorflow"),
   python_version = NULL
 )
@@ -108,16 +108,12 @@ commonly by setting an environment variable:
 }
 
 \section{Apple Silicon}{
- Tensorflow on Apple Silicon is not officially
-supported by the Tensorflow maintainers. However Apple has published a
-custom version of Tensorflow compatible with Arm Macs.
-\code{install_tensorflow()} will install the special packages \code{tensorflow-macos}
-and \code{tensorflow-metal} on Arm Macs. See
+ Beginning with Tensorflow version 2.13, the default
+tensorflow package now works on Apple Silicon. See
 \url{https://developer.apple.com/metal/tensorflow-plugin/} for instructions
-on how to do the equivalent manually. Please note that this is an
-experimental build of both Python and Tensorflow, with known issues. In
-particular, certain operations will cause errors, but can often be remedied
-by pinning them to the CPU. For example:
+on how to install older versions of Tensorflow on macOS. Please note that
+not all operations are supported on Arm Mac GPUs. You can work around the
+missing operations by pinning operations to CPU. For example:
 
 \if{html}{\out{<div class="sourceCode R">}}\preformatted{x <- array(runif(64*64), c(1, 64, 64))
 keras::layer_random_rotation(x, .5)  # Error: