Adaptive Email Security
Frequently Asked Questions
-
What types of Al does Proofpoint use?
Proofpoint uses an ensemble of Al approaches to power behavioral, content and threat network analysis.Proofpoint uses an ensemble of Al approaches to power behavioral, content and threat network analysis. Some of these approaches include:
Behavioral Analysis – Proofpoint uses methods like random forest, neural networks and natural language processing to understand an employee’s trusted relationships, the topics and projects they communicate about and what data they handle. Using these methods, Proofpoint analyzes tens of thousands of email events to create a strong foundation for normal vs. anomalous communications.
Content Analysis – Proofpoint uses methods like topic clustering, natural language processing, large language models and OCR to identify unusual text in images, QR codes, URLs, subject headings, body text, and malicious content inside of email attachments for potential threats. Hundreds of data points within each email are used to determine if the email is safe, malicious or needs review.
Threat Network Analysis – To augment Al approaches, Proofpoint includes third-party threat feeds and data from our customers to improve security across the network. Within minutes, Proofpoint can add malicious domains, URLs, IPs, email addresses and file/image hashes from any source to prevent email-based attacks.
-
Does Proofpoint use machine learning?
Proofpoint uses machine learning (ML) as a part of the platform’s Al ensemble to make predictions.Proofpoint uses machine learning (ML) as a part of the platform’s Al ensemble to make predictions based on:
Email Metadata – Headers, names, attachments, links and more are extracted for analysis to identify when anomalous content is present in an email.
Email Body – Natural language processing models are used to analyze subjects and body text to predict when unusual content should be classified as malicious.
Historical Email Patterns – Deep learning methods are used to understand an employee’s trusted relationships, the topics and projects they communicate about and what data they handle.
-
How is Proofpoint Al different from other Al tools?
Proofpoint Al provides the broadest range of Al applications for email security use cases.Proofpoint Al provides the broadest range of Al applications for email security use cases, including:
Stopping advanced phishing threats – Al-based defense to stop threats like BEC, exec impersonations, image and QR code attacks, and more.
Preventing misdirected emails and files – Al-based protection and customizable policies to prevent sensitive data loss from misdelivered emails
Preventing email data exfiltration – Al-based protection and customizable policies to automatically prevent insiders from exfiltrating data over email.
Delivering in-the-moment coaching – Leverages Al-identified threat signals and provides them to users through intuitive, actionable warnings.
Detecting compromised accounts – Al-based defense to detect suspicious email activity that indicates an internal account takeover.
-
How long has Proofpoint been using Al?
Since filing our first AI patent more than 20 years ago, Proofpoint has been pioneering innovative applications of AI and ML to provide our customers with industry-first solutions that address the top cybersecurity risks they face today.Since filing our first AI patent more than 20 years ago, Proofpoint has been pioneering innovative applications of AI and ML to provide our customers with industry-first solutions that address the top cybersecurity risks they face today. AI and ML are weaved throughout our product suite to give our customers complete and constantly evolving protection against a wide range of threats.
-
Why does Proofpoint use Al?
Proofpoint uses an ensemble of Al-based approaches because it’s the only way to defend against advanced phishing attacks and protect sensitive data on email.Proofpoint uses an ensemble of Al-based approaches because it’s the only way to defend against advanced phishing attacks and protect sensitive data on email. Through behavioral, content and threat network analysis, Proofpoint Al automatically adjusts as our customers hire, acquire and adapt.
-
How does Proofpoint Al help future proof our customers?
Proofpoint Al ensures customers don’t need to know about every new type of bypass and exploit, as malicious emails are generally flagged for a variety of reasons.Proofpoint Al ensures customers don’t need to know about every new type of bypass and exploit, as malicious emails are generally flagged for a variety of reasons. With behavioral and content analysis, Proofpoint Al discovers new exploits for which our threat engineering can subsequently build detection heuristics.
Al Privacy and Security
Frequently Asked Questions
-
Does Proofpoint use any AI or ML products like OpenAI and ChatGPT?
Proofpoint does not use any commercial or third-party-provided AI or ML products that are similar to OpenAI/ChatGPT.Proofpoint does not use any commercial or third-party-provided AI or ML products that are similar to OpenAI/ChatGPT. Our product is built on top of internally developed ML that takes advantage of data science capabilities enabled by services provided within AWS, such as AWS Sagemaker and similar tooling. We do use open source models from HuggingFace in the development of the ML models we own and operate on our platform.
-
Is my company’s data shared with third-party Al or ML products?
Proofpoint exclusively uses customer data within its own systems, and does not use, transfer data, or otherwise integrate with third-party AI or ML products such as ChatGPT/OpenAI, etc.Proofpoint exclusively uses customer data within its own systems, and does not use, transfer data, or otherwise integrate with third-party AI or ML products such as ChatGPT/OpenAI, etc.
-
How does Proofpoint Al use my company’s data?
Proofpoint uses customer data to enhance algorithms and threat network efficacy.Proofpoint uses customer data to enhance algorithms and threat network efficacy. By training the platform’s models on authentic email threat data, Proofpoint can better predict the likelihood of a malicious email, misdirected email or email exfiltration event across our network of customers.
-
How does Proofpoint ensure compliance with legislation like the EU AI Act?
Our legal team and outside counsel are monitoring developments in global Al regulation.Our legal team and outside counsel are monitoring developments in global Al regulation. With regards to the EU Al Act, we expect to fall within the “limited risk” category, given the nature of our services. We note that the EU Al Act will have a 2-year implementation period and its rules will come into effect in late 2026 at the earliest.
-
How does Proofpoint address fundamental Al issues like bias and hallucination?
For hallucinations, we don’t use prompt-engineering to make predictions, but rather fine-tuned models for specific tasks, so hallucinations are not a concern for Proofpoint.For hallucinations, we don’t use prompt-engineering to make predictions, but rather fine-tuned models for specific tasks, so hallucinations are not a concern for Proofpoint.
In terms of bias, we train our data on our randomly selected dataset drawn from a wide customer base, including internal and external emails and from all ages and genders and backgrounds. For identifying malicious emails we build models that are trained on confirmed malicious attacks, and therefore are only biased towards the type of language used in attacks and not towards general types of speech.
In addition, we continually label a portion of the specific cases that are raised as potentially incorrect, allowing us to correct any possible bias that might have come up.
© 2025. All rights reserved. The content on this site is intended for informational purposes only.
Last updated January 30, 2025.