diff --git a/bindings/java/src/lib.rs b/bindings/java/src/lib.rs
index fa225946..1149c752 100644
--- a/bindings/java/src/lib.rs
+++ b/bindings/java/src/lib.rs
@@ -148,7 +148,7 @@ pub unsafe extern "system" fn Java_com_github_null8626_decancer_CuredString_find
           JValueGen::Object(
             &jni_unwrap!(
               env,
-              env.new_string(unsafe { (*inner).get_unchecked(result) })
+              env.new_string(unsafe { &(*inner)[result] })
             )
             .into()
           ),
@@ -188,7 +188,7 @@ pub unsafe extern "system" fn Java_com_github_null8626_decancer_CuredString_find
           JValueGen::Object(
             &jni_unwrap!(
               env,
-              env.new_string(unsafe { (*inner).get_unchecked(result) })
+              env.new_string(unsafe { &(*inner)[result] })
             )
             .into()
           ),
diff --git a/bindings/native/src/lib.rs b/bindings/native/src/lib.rs
index 9e17652f..d458f912 100644
--- a/bindings/native/src/lib.rs
+++ b/bindings/native/src/lib.rs
@@ -378,7 +378,7 @@ pub unsafe extern "C" fn decancer_equals_wide(
   other_size: usize,
 ) -> bool {
   utf16::get(other_str, other_size)
-    .map(|vec| unsafe { (*cured) == str::from_utf8_unchecked(&vec) })
+    .map(|vec| unsafe { (*cured) == str::from_utf8(&vec).unwrap() })
     .unwrap_or_default()
 }
 
@@ -403,7 +403,7 @@ macro_rules! comparison_fn {
         other_size: usize,
       ) -> bool {
         utf16::get(other_str, other_size)
-          .map(|vec| unsafe { (*cured).$name(str::from_utf8_unchecked(&vec)) })
+          .map(|vec| unsafe { (*cured).$name(str::from_utf8(&vec).unwrap()) })
           .unwrap_or_default()
       }
     }
diff --git a/bindings/native/src/utf16.rs b/bindings/native/src/utf16.rs
index 3d218722..988f5415 100644
--- a/bindings/native/src/utf16.rs
+++ b/bindings/native/src/utf16.rs
@@ -64,7 +64,7 @@ pub(crate) unsafe fn get_array(
     output.push(unsafe {
       let s = input_ptr.offset(i as _);
 
-      String::from_utf8_unchecked(get((*s).string, (*s).size)?)
+      String::from_utf8(get((*s).string, (*s).size)?).unwrap()
     });
   }
 
diff --git a/bindings/native/src/utf8.rs b/bindings/native/src/utf8.rs
index 51fbdb9b..704a1cbb 100644
--- a/bindings/native/src/utf8.rs
+++ b/bindings/native/src/utf8.rs
@@ -21,11 +21,9 @@ pub(crate) fn get(input_ptr: *mut u8, mut input_size: usize) -> Option<&'static
     input_size = input_ptr.size;
   }
 
-  unsafe {
-    Some(str::from_utf8_unchecked(slice::from_raw_parts(
-      input_ptr, input_size,
-    )))
-  }
+  str::from_utf8(unsafe {slice::from_raw_parts(
+    input_ptr, input_size,
+  )}).ok()
 }
 
 pub(crate) unsafe fn get_array(
diff --git a/bindings/node/src/lib.rs b/bindings/node/src/lib.rs
index fc2ff89e..957f0a4f 100644
--- a/bindings/node/src/lib.rs
+++ b/bindings/node/src/lib.rs
@@ -7,7 +7,7 @@ use napi::{
   bindgen_prelude::{Error, FromNapiValue},
   Env, JsNumber, JsString, JsUnknown, Result, Status, ValueType,
 };
-use std::{mem::transmute, ops::Range};
+use std::ops::Range;
 
 macro_rules! options {
   (
@@ -116,7 +116,7 @@ impl CuredString {
   fn new_match(&self, mat: Range<usize>) -> Match {
     Match {
       range: mat.clone(),
-      portion: String::from(unsafe { self.0.get_unchecked(mat) }),
+      portion: String::from(&self.0[mat]),
     }
   }
 
@@ -218,7 +218,7 @@ fn cure(input: String, maybe_options: JsUnknown) -> Result<CuredString> {
     <Option<Options> as FromNapiValue>::from_unknown(maybe_options).map(options)
   }?;
 
-  match decancer::cure(&input, unsafe { transmute(options) }) {
+  match decancer::cure(&input, options.into()) {
     Ok(output) => Ok(CuredString(output)),
     Err(err) => Err(Error::new(Status::InvalidArg, err)),
   }
diff --git a/core/src/bidi/paragraph.rs b/core/src/bidi/paragraph.rs
index 021859ef..0ba68931 100644
--- a/core/src/bidi/paragraph.rs
+++ b/core/src/bidi/paragraph.rs
@@ -575,24 +575,27 @@ impl Paragraph {
             last.level.new_explicit_next_ltr()
           };
 
-          if new_level.is_ok() && overflow_isolate_count == 0 && overflow_embedding_count == 0 {
-            // SAFETY: new_level was already proven to be Ok
-            let new_level = unsafe { new_level.unwrap_unchecked() };
-
-            stack.push(Status {
-              level: new_level,
-              status: original_classes[idx].override_status(),
-            });
+          match (new_level, overflow_isolate_count, overflow_embedding_count) {
+            (Ok(new_level), 0, 0) => {
+              stack.push(Status {
+                level: new_level,
+                status: original_classes[idx].override_status(),
+              });
+
+              if is_isolate {
+                valid_isolate_count += 1;
+              } else {
+                levels[idx] = new_level;
+              }
+            }
 
-            if is_isolate {
-              valid_isolate_count += 1;
-            } else {
-              levels[idx] = new_level;
+            _ => {
+              if is_isolate {
+                overflow_isolate_count += 1;
+              } else if overflow_isolate_count == 0 {
+                overflow_embedding_count += 1;
+              }
             }
-          } else if is_isolate {
-            overflow_isolate_count += 1;
-          } else if overflow_isolate_count == 0 {
-            overflow_embedding_count += 1;
           }
 
           if !is_isolate {
@@ -700,8 +703,7 @@ impl Paragraph {
         .unwrap_or(start_class);
 
       let mut sequence = if start_class == Class::PDI && stack.len() > 1 {
-        // SAFETY: stack is already checked to be not empty
-        unsafe { stack.pop().unwrap_unchecked() }
+        stack.pop().unwrap()
       } else {
         Vec::with_capacity(1)
       };
diff --git a/core/src/leetspeak.rs b/core/src/leetspeak.rs
index b1615a92..519b37b9 100644
--- a/core/src/leetspeak.rs
+++ b/core/src/leetspeak.rs
@@ -52,7 +52,7 @@ pub(crate) fn find(haystack: &str, character: u32) -> Option<usize> {
     _ => return None,
   };
 
-  let regex = unsafe { REGEXES.as_slice().get_unchecked(idx as usize).as_ref() };
+  let regex = &REGEXES[idx as usize];
 
-  regex?.find(haystack).map(|mat| mat.len())
+  regex.as_ref()?.find(haystack).map(|mat| mat.len())
 }
diff --git a/core/src/lib.rs b/core/src/lib.rs
index b31ffcc3..3b87e95f 100644
--- a/core/src/lib.rs
+++ b/core/src/lib.rs
@@ -89,14 +89,9 @@ const fn is_special_rtl(code: u32) -> bool {
 }
 
 fn cure_char_inner(code: u32, options: Options) -> Translation {
-  // SAFETY: even if there is no lowercase mapping for some codepoints, it would just return itself.
-  // therefore, the first iteration and/or codepoint always exists.
-  let code_lowercased = unsafe {
-    char::from_u32_unchecked(code)
-      .to_lowercase()
-      .next()
-      .unwrap_unchecked() as _
-  };
+  let code_lowercased = char::from_u32(code)
+    .and_then(|character| character.to_lowercase().next())
+    .unwrap() as _;
 
   let is_case_sensitive = code != code_lowercased;
 
@@ -281,8 +276,7 @@ fn first_cure_pass(input: &str) -> (String, Vec<Class>, Vec<Paragraph>) {
           _ => {}
         }
 
-        // SAFETY: the only modification to this codepoint is in the if-statement above.
-        refined_input.push(unsafe { char::from_u32_unchecked(codepoint) });
+        refined_input.push(char::from_u32(codepoint).unwrap());
 
         idx += character_len;
       }
@@ -438,11 +432,8 @@ macro_rules! cure {
 #[macro_export]
 macro_rules! format {
   ($string:expr) => {
-    // SAFETY: having disable_bidi enabled removes all possibilities of an Err
-    unsafe {
-      $crate::cure($string, $crate::Options::formatter())
-        .unwrap_unchecked()
-        .into()
-    }
+    $crate::cure($string, $crate::Options::formatter())
+      .unwrap()
+      .into()
   };
 }
diff --git a/core/src/similar.rs b/core/src/similar.rs
index 99e2087d..fee10352 100644
--- a/core/src/similar.rs
+++ b/core/src/similar.rs
@@ -10,10 +10,8 @@ pub(crate) const SIMILAR_START: u16 = read_u16_le(unsafe { CODEPOINTS.offset(2)
 pub(crate) const SIMILAR_END: u16 = read_u16_le(unsafe { CODEPOINTS.offset(4) });
 
 pub(crate) fn is(self_char: char, other_char: char) -> bool {
-  // SAFETY: even if there is no lowercase mapping for some codepoints, it would just return itself.
-  // therefore, the first iteration and/or codepoint always exists.
-  let self_char = unsafe { self_char.to_lowercase().next().unwrap_unchecked() as u32 };
-  let other_char = unsafe { other_char.to_lowercase().next().unwrap_unchecked() as u32 };
+  let self_char = self_char.to_lowercase().next().unwrap() as u32;
+  let other_char = other_char.to_lowercase().next().unwrap() as u32;
 
   if self_char == other_char {
     return true;
@@ -93,8 +91,7 @@ impl<'a> Iterator for CachedPeek<'a> {
       return None;
     }
 
-    // SAFETY: the current index always points to an existing element
-    let current = unsafe { *self.cache.get_unchecked(self.index) };
+    let current = self.cache[self.index];
     let next_element = self.next_value();
 
     if next_element.is_none() {
@@ -124,26 +121,21 @@ impl<'a, 'b> Matcher<'a, 'b> {
       self_str = "";
     }
 
-    // SAFETY: self_str = "" would immediately cancel out the first iteration.
     Self {
       self_iterator: self_str.chars(),
       #[cfg(feature = "leetspeak")]
       self_str,
       self_index: 0,
-      other_iterator: CachedPeek::new(other_chars, unsafe { other_first.unwrap_unchecked() }),
+      other_iterator: CachedPeek::new(other_chars, other_first.unwrap()),
     }
   }
 
   #[cfg(feature = "leetspeak")]
   fn matches_leetspeak(&mut self, other_char: char) -> Option<usize> {
-    // SAFETY: already guaranteed to be within the string's bounds.
-    let haystack = unsafe { self.self_str.get_unchecked(self.self_index..) };
+    let haystack = &self.self_str[self.self_index..];
     let matched_len = leetspeak::find(haystack, other_char as _)?;
 
-    // SAFETY: this will never go out of bounds as well
-    //         the furthest it would go would be an empty string.
-    self.self_iterator =
-      unsafe { self.self_str.get_unchecked(self.self_index + matched_len..) }.chars();
+    self.self_iterator = self.self_str[self.self_index + matched_len..].chars();
 
     Some(matched_len)
   }
@@ -241,8 +233,7 @@ impl<'a, 'b> Iterator for Matcher<'a, 'b> {
               self.other_iterator.restart();
 
               current_separator = None;
-              // SAFETY: this state of the program wouldn't be accessible if the first iteration returns a None
-              current_other = unsafe { self.other_iterator.next().unwrap_unchecked() };
+              current_other = self.other_iterator.next().unwrap();
 
               let (skipped, matched_skip) = self.skip_until(current_other.0)?;
 
diff --git a/core/src/translation.rs b/core/src/translation.rs
index fed434fb..98a82d94 100644
--- a/core/src/translation.rs
+++ b/core/src/translation.rs
@@ -47,7 +47,7 @@ impl Translation {
   #[cfg(feature = "options")]
   pub(crate) fn into_uppercase(self) -> Self {
     match self {
-      Self::Character(c) => Self::Character(unsafe { c.to_uppercase().next().unwrap_unchecked() }),
+      Self::Character(c) => Self::Character(c.to_uppercase().next().unwrap()),
       Self::String(s) => Self::String(Cow::Owned(s.as_ref().to_uppercase())),
       Self::None => Self::None,
     }
diff --git a/core/src/util.rs b/core/src/util.rs
index 12ce599c..b9444191 100644
--- a/core/src/util.rs
+++ b/core/src/util.rs
@@ -44,8 +44,8 @@ where
   let mut j = 0;
 
   for i in 1..ranges.len() {
-    let current = unsafe { ranges.get_unchecked(i).clone() };
-    let previous = unsafe { ranges.get_unchecked_mut(j) };
+    let current = ranges[i].clone();
+    let previous = &mut ranges[j];
 
     if current.start >= previous.start && current.start <= previous.end {
       previous.end = max(current.end, previous.end);