Instantly know what's human and AI on Twitter, LinkedIn, Substack and more. Get our new chrome extension.

The job market has changed. Candidates are using generative AI tools to mass-produce job applications. Some of the most popular generative AI tools are ChatGPT and Claude. Candidates are also using dedicated resume builders like Teal and Rezi.
Generative AI tools promise efficiency. However, they have flooded HR teams with low-quality, undifferentiated applications. These applications are clogging hiring pipelines.
To combat this issue, modern Applicant Tracking Systems (ATS) are evolving. They aren't just parsing for keywords. Instead, they are using enterprise-grade verification tools, like Pangram, as a diagnostic layer. This guide explores:
Applicant Tracking Systems (ATS) automatically parse, sort, and rank incoming application based on job-description keywords. Now, many ATS systems are also incorporating AI detection APIs that let them automatically filter out AI-generated applications.
Historically, applicants tried to beat the ATS by stuffing their resumes with keywords. This used to work, but modern ATS platforms have grown far more sophisticated. Now, they prioritize semantic context and authentic experience over keyword density.
ATS software is primarily focused on simple sorting. Due to the rise of “Auto-Apply” bots, ATS software is shifting to provenance verification. Being able to verify the provenance of a resume lets recruiters see if a real human actually produced it, allowing them to make the best possible hiring decision.
Can recruiters detect AI-generated resumes? Yes: most recruiters use AI resume checker tools to analyze an application's structural predictability. These resume checker tools easily flag documents that contain a lot of LLM generated text.
Resumes are highly structured documents. They are inherently formal. This is why AI can generate them with some measure of ease. But AI models leave distinct statistical fingerprints when they generate documents. They often rely on hyper-predictable formatting patterns. Also, they often produce sentences that are exactly the same length/size.
If you are using AI to generate resumes, keep in mind that AI detectors don't look for “bad writing.” They look for the mathematical signatures inherent to AI-generated text. You may not notice these signatures, but resume checker AI tools like Pangram do notice it. Recruiters use these resume checker AI tools.
Over-relying on an AI resume builder strips an application of its unique human voice and the specific anecdotes that comprise it. The absence of these two things leads to an average, completely undifferentiated candidate. An average, undifferentiated candidate will not stand out to recruiters.
LLMs are designed to produce the most statistically likely response. Not the “best” response. Not the “most useful” response. The most statistically likely response. Because of this, an AI-generated resume will look like thousands of other AI-generated resumes.
Pangram has produced a body of research on AI phrases. AI resume builders tend to overuse words like “spearheaded,” “orchestrated,” and/or “synergized.” These are words that rarely reflect how real professionals speak during an interview. Seeing these words in a resume can make a recruiter think that it was AI-generated.
Looking for statistically likely responses is how an AI resume checker is able to detect AI generated resumes. Going over the AI phrases in a resume supports these detection efforts. If you want your resume to stand out, both of these AI tells will hinder you.
Rather than auto-rejecting candidates, HR teams integrate Pangram's API into their enterprise workflow. Pangram allows them to distinguish between the different types of AI usage. There is a difference between a candidate who used AI for light proofreading and a candidate that used AI to generate their entire resume. Pangram can distinguish between these candidates and perform a resume review to detect AI.
Recruiters use Pangram to improve the signal-to-noise ratio in their applicant pools. A resume with heavy AI usage can be dismissed and candidates with authentic applications can be prioritized.
If an HR team is ever accused of unfair labor practices, they will need data for their defense. Pangram has an industry-leading accuracy rate of 99.98% and a low false-positive rate. Other AI detectors can have high false positive rates for text written by ESL authors, but Pangram performs well on ESL text.
Resume builders can be worth it, but only in specific contexts. If you need help with high-level structuring and/or making sure your grammar is accurate, AI resume builders are worth it. However, fully outsourcing your resume creation to an AI will likely result in your application being flagged and rejected.
If you would like to use generative AI to build your resume, think of it as a “brainstorming assistant.” Use it for ideas and guidance. Don’t use it as a “ghostwriter,” though, because your application is likely to be flagged.
Even if an AI-generated resume passes an ATS AI detection, the bullet points in your resume will be addressed during a live interview. If these bullet points were hallucinated by AI - or if any other facet of your application is inaccurate - the truth will come out.
The hiring process is becoming a tech-driven arms race. Candidates are scaling output with AI. HR teams are scaling verification with AI detection.
The ultimate competitive advantage in 2026 isn't a flawlessly robotic resume. It is demonstrable human authenticity.
Ensure your hiring pipeline is filled with authentic talent, not automated spam. Integrate Pangram’s API into your ATS today.






