diff --git a/.github/scripts/gen-changelog.py b/.github/scripts/gen-changelog.py
new file mode 100755
index 0000000000..cccb04be91
--- /dev/null
+++ b/.github/scripts/gen-changelog.py
@@ -0,0 +1,335 @@
+#!/usr/bin/env python3
+"""
+Generate changelog for xemu releases.
+
+This script generates a markdown file that includes:
+- Pull requests merged since the last release
+- Issues closed since the last release
+- Affected game titles (extracted from issue bodies)
+
+Requires the GitHub CLI (gh) to be installed and authenticated.
+"""
+
+import argparse
+import json
+import re
+import subprocess
+import sys
+from collections import defaultdict
+from urllib.request import urlopen
+from urllib.error import URLError
+
+
+REPO_NAME = "xemu-project/xemu"
+XEMU_TITLE_URL_BASE = "https://xemu.app/titles/"
+XDB_RAW_URL_BASE = "https://raw.githubusercontent.com/xemu-project/xdb/main/titles"
+
+
+def get_title_name(title_id: str) -> str | None:
+    """
+    Fetch title name from xemu-project/xdb for a specific title ID.
+    Title IDs are 8 hex chars: first 2 bytes are publisher code (ASCII), last 2 bytes are title number.
+    """
+    title_id = title_id.lower()
+    if len(title_id) != 8:
+        return None
+
+    try:
+        pub_code = bytes.fromhex(title_id[:4]).decode("ascii")
+        title_num = int(title_id[4:], 16)
+    except (ValueError, UnicodeDecodeError):
+        return None
+
+    info_url = f"{XDB_RAW_URL_BASE}/{pub_code}/{title_num:03d}/info.json"
+    try:
+        with urlopen(info_url, timeout=5) as response:
+            info = json.loads(response.read().decode("utf-8"))
+            return info.get("name")
+    except (URLError, json.JSONDecodeError, TimeoutError):
+        return None
+
+
+def get_previous_tag(current_ref: str) -> str | None:
+    """Get the previous release tag before the current ref."""
+    try:
+        result = subprocess.run(
+            [
+                "git",
+                "describe",
+                "--tags",
+                "--match",
+                "v*",
+                "--abbrev=0",
+                f"{current_ref}^",
+            ],
+            capture_output=True,
+            text=True,
+            check=True,
+        )
+        return result.stdout.strip()
+    except subprocess.CalledProcessError:
+        return None
+
+
+def get_commits_between(from_ref: str, to_ref: str) -> list[str]:
+    """Get list of commit SHAs between two refs."""
+    try:
+        result = subprocess.run(
+            [
+                "git",
+                "log",
+                "--pretty=format:%H",
+                "--first-parent",
+                f"{from_ref}..{to_ref}",
+            ],
+            capture_output=True,
+            text=True,
+            check=True,
+        )
+        if result.stdout.strip():
+            return result.stdout.strip().split("\n")
+        return []
+    except subprocess.CalledProcessError:
+        return []
+
+
+def extract_title_ids(text: str) -> set[str]:
+    """Extract Xbox title IDs from issue body text."""
+    if not text:
+        return set()
+    titles_re = re.compile(r"Titles?[:/]\s*([a-fA-F0-9,\s]+)", re.IGNORECASE)
+    title_id_re = re.compile(r"([a-fA-F0-9]{8})")
+    references = " ".join(titles_re.findall(text))
+    return set(title_id_re.findall(references.lower()))
+
+
+def format_title_reference(title_id: str, title_cache: dict[str, str | None]) -> str:
+    """Format a title ID as a markdown link with the game name."""
+    title_id = title_id.lower()
+    if title_id not in title_cache:
+        title_cache[title_id] = get_title_name(title_id)
+    name = title_cache[title_id]
+    if name:
+        return f"[{name}]({XEMU_TITLE_URL_BASE}{title_id})"
+    return f"`{title_id}`"
+
+
+def log(msg: str) -> None:
+    """Print a log message to stderr."""
+    print(f"[changelog] {msg}", file=sys.stderr)
+
+
+def gh_api(endpoint: str) -> dict | list | None:
+    """Call GitHub REST API using gh CLI."""
+    try:
+        result = subprocess.run(
+            ["gh", "api", f"repos/{REPO_NAME}/{endpoint}"],
+            capture_output=True,
+            text=True,
+            check=True,
+        )
+        return json.loads(result.stdout)
+    except (subprocess.CalledProcessError, json.JSONDecodeError):
+        return None
+
+
+def gh_graphql(query: str) -> dict | None:
+    """Call GitHub GraphQL API using gh CLI."""
+    try:
+        result = subprocess.run(
+            ["gh", "api", "graphql", "-f", f"query={query}"],
+            capture_output=True,
+            text=True,
+            check=True,
+        )
+        return json.loads(result.stdout)
+    except (subprocess.CalledProcessError, json.JSONDecodeError):
+        return None
+
+
+def generate_changelog(
+    current_ref: str,
+    previous_tag: str | None = None,
+) -> str:
+    """Generate changelog markdown content."""
+    title_cache: dict[str, str | None] = {}
+
+    if previous_tag is None:
+        log(f"Finding previous tag before {current_ref}...")
+        previous_tag = get_previous_tag(current_ref)
+
+    if previous_tag is None:
+        return "No previous release found.\n"
+
+    log(f"Generating changelog from {previous_tag} to {current_ref}")
+
+    commits = get_commits_between(previous_tag, current_ref)
+    commit_shas = set(commits)
+    log(f"Found {len(commit_shas)} commits in range")
+    if not commit_shas:
+        return "No changes since last release.\n"
+
+    log("Fetching pull requests for commits...")
+    merged_prs: list[dict] = []
+    pr_numbers_seen: set[int] = set()
+
+    for sha in commit_shas:
+        pulls = gh_api(f"commits/{sha}/pulls")
+        if not pulls:
+            continue
+        for pr in pulls:
+            pr_num = pr["number"]
+            if pr_num in pr_numbers_seen:
+                continue
+            pr_numbers_seen.add(pr_num)
+
+            # Skip dependabot PRs that only touch CI/workflow files
+            if pr["user"]["login"] == "dependabot[bot]":
+                files = gh_api(f"pulls/{pr_num}/files") or []
+                if all(f["filename"].startswith(".github/") for f in files):
+                    log(f"  Skipping CI-only dependabot PR #{pr_num}: {pr['title']}")
+                    continue
+
+            merged_prs.append(pr)
+            log(f"  Found PR #{pr_num}: {pr['title']}")
+
+    log(f"Found {len(merged_prs)} merged PRs")
+    merged_prs.sort(key=lambda pr: pr.get("merged_at") or "", reverse=True)
+
+    log("Fetching closed issues via GraphQL...")
+    closed_issues: list[dict] = []
+    issue_numbers_seen: set[int] = set()
+    affected_titles: dict[str, list[dict]] = defaultdict(list)
+
+    if merged_prs:
+        owner, name = REPO_NAME.split("/")
+        pr_numbers = [pr["number"] for pr in merged_prs]
+        pr_queries = " ".join(
+            f"pr{pr_num}: pullRequest(number: {pr_num}) {{ closingIssuesReferences(first: 10) {{ nodes {{ number title body state stateReason }} }} }}"
+            for pr_num in pr_numbers
+        )
+        query = f'{{ repository(owner: "{owner}", name: "{name}") {{ {pr_queries} }} }}'
+
+        result = gh_graphql(query)
+        if result:
+            repo_data = result.get("data", {}).get("repository", {})
+            for pr_num in pr_numbers:
+                pr_data = repo_data.get(f"pr{pr_num}", {})
+                issues = pr_data.get("closingIssuesReferences", {}).get("nodes", [])
+                for issue in issues:
+                    if issue["number"] in issue_numbers_seen:
+                        continue
+                    if (
+                        issue["state"] == "CLOSED"
+                        and issue["stateReason"] == "COMPLETED"
+                    ):
+                        issue_numbers_seen.add(issue["number"])
+                        closed_issues.append(issue)
+                        log(f"  Found issue #{issue['number']}: {issue['title']}")
+                        title_ids = extract_title_ids(issue.get("body") or "")
+                        for tid in title_ids:
+                            affected_titles[tid].append(issue)
+
+    log(f"Found {len(closed_issues)} closed issues")
+    if affected_titles:
+        log(f"Fetching names for {len(affected_titles)} affected titles...")
+        for tid in affected_titles.keys():
+            if tid not in title_cache:
+                title_cache[tid] = get_title_name(tid)
+                name = title_cache[tid]
+                if name:
+                    log(f"  {tid} -> {name}")
+                else:
+                    log(f"  {tid} -> (not found)")
+
+    log("Building changelog...")
+    lines = []
+
+    if merged_prs:
+        lines.append("## Pull Requests")
+        lines.append("")
+        for pr in merged_prs:
+            lines.append(f"* #{pr['number']} - {pr['title']} (@{pr['user']['login']})")
+        lines.append("")
+
+    if closed_issues:
+        lines.append("## Issues Fixed")
+        lines.append("")
+        for issue in closed_issues:
+            lines.append(f"* #{issue['number']} - {issue['title']}")
+        lines.append("")
+
+    if affected_titles:
+        lines.append("## Affected Titles")
+        lines.append("")
+        lines.append("The following titles had issues fixed in this release:")
+        lines.append("")
+        for tid in sorted(affected_titles.keys()):
+            issue_nums = ", ".join(f"#{i['number']}" for i in affected_titles[tid])
+            title_ref = format_title_reference(tid, title_cache)
+            lines.append(f"* {title_ref} ({issue_nums})")
+        lines.append("")
+
+    if not merged_prs and not closed_issues:
+        log("No PRs or issues found, falling back to git log")
+        lines.append("## Changes")
+        lines.append("")
+        try:
+            result = subprocess.run(
+                [
+                    "git",
+                    "log",
+                    "--pretty=format:* %h %s",
+                    "--first-parent",
+                    f"{previous_tag}..{current_ref}",
+                ],
+                capture_output=True,
+                text=True,
+                check=True,
+            )
+            lines.append(result.stdout)
+        except subprocess.CalledProcessError:
+            lines.append("See commit history for changes.")
+        lines.append("")
+
+    compare_url = (
+        f"https://github.com/{REPO_NAME}/compare/{previous_tag}...{current_ref}"
+    )
+    lines.append(f"**Full Changelog**: {compare_url}")
+
+    log("Done")
+    return "\n".join(lines)
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Generate changelog for xemu releases")
+    parser.add_argument(
+        "--ref",
+        default="HEAD",
+        help="Git ref to generate changelog for (default: HEAD)",
+    )
+    parser.add_argument(
+        "--previous-tag",
+        help="Previous release tag to compare against (auto-detected if not specified)",
+    )
+    parser.add_argument(
+        "--output",
+        "-o",
+        help="Output file (default: stdout)",
+    )
+
+    args = parser.parse_args()
+    changelog = generate_changelog(
+        current_ref=args.ref,
+        previous_tag=args.previous_tag,
+    )
+    if args.output:
+        with open(args.output, "w") as f:
+            f.write(changelog)
+        print(f"Changelog written to {args.output}", file=sys.stderr)
+    else:
+        print(changelog)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index bdb4e5757c..30022460f2 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -41,13 +41,10 @@ jobs:
       with:
         fetch-depth: 0
         ref: ${{ inputs.ref }}
-    - name: Generate Changelog
-      run: |
-        if PREVIOUS_TAG=$(git describe --tags --match 'v*' --abbrev=0 HEAD^ 2>/dev/null); then
-          git log --pretty=format:'* %h %s' --first-parent $PREVIOUS_TAG..HEAD > CHANGELOG.md
-        else
-          touch CHANGELOG.md
-        fi
+    - name: Generate release notes
+      env:
+        GH_TOKEN: ${{ github.token }}
+      run: python3 .github/scripts/gen-changelog.py --ref ${{ inputs.ref }} -o CHANGELOG.md
     - name: Download artifacts
       uses: actions/download-artifact@37930b1c2abaa49bbe596cd826c3c89aef350131 # v4
       with: