How to Tell if an Email Was Written by AI

Security Risks

• •

  Sophisticated phishing attempts:

  Cybercriminals increasingly use AI to craft realistic phishing emails. Using AI, these emails appear genuine and are highly personalized. This makes it more challenging than ever to detect, presenting email users with serious security concerns.

   

• •

  Advanced spam techniques:

  AI allows spammers to bypass filters by mimicking human language patterns, reducing the effectiveness of traditional anti-spam measures. These emails might seem more sophisticated, avoiding common red flags like poor grammar, spelling mistakes, and awkward phrasing.

   

Marketing & communication quality

• •

  Lack of personalization and emotional nuance:

  While AI can personalize emails to some extent, they often need more depth and empathy than human-crafted messages. Customers usually sense whether an email lacks authenticity, which may affect their trust and engagement with your brand.

   

• •

  Potential for inaccuracies:

  AI often lacks contextual understanding or adherence to brand guidelines, which can result in messaging that doesn’t align with your brand’s voice. While AI tends to apply a “one-size-fits-all” approach, people have unique preferences and cultural nuances, which the AI may overlook or misinterpret.

   

• •

  Importance of the human touch in customer support:

  Automated responses are great for quick and easy answers, but they might need help to handle complicated or personal problems. A natural person can build a better relationship with your customers and show them you care. They can also understand things that AI might miss.

   

3 primary signs of AI authorship

1. The writing style

   AI-generated emails often follow a set pattern, like a robot’s checklist. They might sound too perfect without the little things that make human writing feel natural, like changing sentence length or using everyday language.

    

   Look for emails that seem too formal or stiff; you might be talking to a bot.

    

2. Repetitive phrasing

   AI emails might say the same thing repeatedly, even in the same or different emails. This is because AI learns from what it’s seen before and sometimes gets stuck on the exact phrases or sentences.

    

   For example, it might repeat the same catchphrase, use the same greeting or goodbye too many times, or say the same thing almost the same way.

    

3. Contextual irrelevance

   AI is good at mimicking human speech, but it sometimes misses the big picture. For example, an AI email might say something that doesn’t fit the situation, what you like, or what you were talking about before.

    

   Look for emails that seem weird or need clarification, like suggestions that don’t fit the situation or extra details that don’t add anything to the conversation.

    

3 secondary signs of AI authorship

1. Source metadata

   Behind every email lies metadata that tells you where and when they were sent. AI emails might have weird timestamps, IP addresses from unexpected places, or many emails sent fast.

    

   If an email is from an AI, check its metadata. You might find some clues that show it’s not from a real person.

    

2. Email headers

   Email headers are like a map that shows where an email came from and where it went. AI emails might have weird things in the header, like a sender’s address that doesn’t match the name or a strange path the email took.

    

   Look for funny sender names, weird IP addresses, or if the sender’s address doesn’t match the domain they claim to be from.

    

3. Engagement metrics

   AI emails often follow a schedule, like a robot’s alarm clock. They might be sent at weird times or in big batches. If you notice emails from someone always arrive simultaneously or in huge numbers, it might be an AI.

    

   You can also check how people react to these emails. If they aren’t opened often, clicked on rarely, or replied to rarely, it might be an AI.

    

    

3 tertiary signs of AI authorship

1. Response patterns

   AI emails can reply super fast, like a robot answering a question. If someone replies to your email right away or at weird times, it might be an AI. Look for patterns of quick and consistent responses.

    

2. Comparison

   Compare AI emails to ones you know are from real people. Look for differences in how they sound and feel and how complicated they are.

    

   If you notice a change, like a person suddenly sounding too formal or using the exact phrases a lot, it might be an AI.

    

3. Recipient targeting

   AI can send emails that look just for you, but they might miss the small details that make a natural person feel special.

    

   Look for emails that call you by name but don’t know much about you, use generic language, or don’t make sense to you. These might be AI emails that were sent to a lot of people at once.

    

1. OrginalityAI:

   OriginalityAI uses machine learning to analyze the writing style and patterns typical of AI-authored text. It then provides you an ‘AI likelihood’ score to show how likely it is to be from an AI. But it might have trouble with well-written emails.

    

2. CopyLeaks:

   CopyLeaks is a popular tool that tells you how likely it is that an AI wrote an email. It gives you a percentage score. Although pretty robust, its results are not foolproof, particularly with content that blends AI and human-written sections.

    

3. GPTZero:

   GPTZero is designed to detect content written by GPT models (e.g., ChatGPT). It looks at the words used and how the sentences are combined to see if the content is from an AI. It works well for emails from this AI but might fail by newer or smarter AIs.

    

1. Read the email carefully.

   Does it feel like a real person wrote it, or does it sound stiff and robotic? AI-written emails often follow a set pattern, like a checklist. They might jump from one topic to another without making sense, or the sentences might feel like they need to belong together.

    

   These are signs that an AI might have written the email.

    

2. Check for repetition

   Pay attention to whether the email repeatedly uses the exact words or phrases. AI often repeats things because it learns from what it’s seen before.

    

   Humans change how they write, but AI might get stuck on certain words or phrases.

    

3. Analyze the tone

   Does the email feel like a real person wrote it? AI emails might be too severe or formal or sound different from the person you expect.

    

   If the email uses your name but doesn’t know much about you, or if it uses the exact old phrases, it might be an AI.