safeDM

• 

  Spot creepy DMs instantly with SafeDM

Inspiration

As a girl on the internet, I’ve dealt with DMs that felt uncomfortable, inappropriate, or just “off.” My friends have too. SafeDM was born from that shared experience — the desire for something gentle that quietly says: “This message isn’t okay, and you deserve to feel safe.”

What it does

SafeDM flags creepy or manipulative DMs in real time, shows the red flags, and offers calm, confident replies to help you protect your peace online. It works on Reddit, Instagram Web, X, and LinkedIn , all fully on device.

How I built it

Using AI-assisted vibe coding, I built a Chrome extension with content scripts, DOM observers, and a soft black–hot pink UI. Everything runs locally to keep users safe and private.

Challenges I ran into

• Extracting message text from complex chat UIs
  
• Injecting UI without breaking site layouts
  
• Building a privacy‑first detector
  
• Making the tool feel supportive, not scary
  

Accomplishments that I'm proud of

Built my first Chrome extension Turned a personal issue into a real tool Made safety feel soft and feminine Fully private, on device detection Learned DOM scripting + UI injection Used AI creatively, not passively Created a working multi-platform prototype

What I learned

I learned that safety tools can be feminine, soft, and empowering and that technology can protect our emotional wellbeing, not just our data.

What's next for safeDM

Add smarter AI detection Support more platforms Custom sensitivity settings Build a wellness dashboard One-click block + report On-device tone analysis Publish on Chrome Web Store

Built With

• ai-assisted
• chatgpt)
• chrome-extensions-(mv3)
• css
• development
• dom-apis
• html
• javascript
• mutation-observers