Today, companies should expect at least 10% of their candidate pipelines to engage in fraudulent behavior. In this article, Dan Finnigan, CEO of Filtered and a veteran leader in hiring tech, provides practical advice on how companies can protect themselves using AI advancements and process changes.Â
Fraud is the dirty little secret in tech hiring that plagues everyone. Over the past decade, and certainly, through the rise of pandemic-driven remote interviewing, it has become ubiquitous. At Filtered, our data consistently show that about 10% of candidates taking pre-hire assessments are flagged for fraud across all our clients. And that is when the candidates know they are being monitored for it! For unchecked pipelines, the percentage is likely much higher.Â
Allowing candidates to cheat on hiring assessments can be enormously costly. If left unchecked, fraud causes tens or hundreds of thousands of dollars of damages every time. Not to mention, it is often traumatic for the whole team.Â
Yet, we have all seen these fake candidate scams, like the plagiarist who can fake the test but sinks the job or the invented person who started all their social accounts last week. Are you doing enough to weed them out of your pipeline before they cost you?Â
If you have not covered the following bases for hiring technical staff, chances are you are vulnerable.
1. Use AI To Identify Location Mismatches
Occasionally, the actual location of a candidate is not the same as where they claim to be when responding to screening questions or taking an online assessment. Sometimes this is acceptable. For instance, if a candidate is generalizing their location by saying “Boston†instead of their exact location of the suburb “Quincy, MA.†However, this should be considered a red flag, particularly if the two locations are in different countries. At this point, hiring managers should probe further to uncover how much more the candidate may be misrepresenting.
2. Use AI To Score Similarity and Detect Plagiarism
For any written response, especially code assessments, sometimes a candidate’s response to a question will be incredibly similar to that of others. Usually, this is also a red flag. When answers are too identical, it can indicate the answer was shared online or with the candidate directly. Candidates with high similarity scores to others require either further testing or closer evaluation to ensure they can do the work by themselves.
3. Monitor and Score Open Tabs During Interviews
We all know it is easy to let open tabs accumulate in your browser. Occasionally, candidates intentionally open tabs before and during the interview for research or reference. This is often what they would do in real life. After all, a developer’s favorite tool is a search engine, and it is not necessarily a bad practice. However, not all searches are equal, and some types can indicate a lack of basic understanding or outright piracy. Detecting all the open tabs during an interview (with the candidate’s consent) and running an inventory analysis can help identify potential misuse. Better yet, allow your interviewers to look at the tabs and visit the sites to better evaluate if the candidate was cheating or just being thoroughly prepared.Â
See More: The Resume is Dying: Here’s What the Future of Hiring Looks Like
4. Count Their Mouse Exits
Building on the similarity comparison and the open tabs, mouse exits can also indicate that a candidate is leaving the interview or testing environment to search for answers. Of course, there are several valid reasons that a candidate’s mouse cursor can leave the testing environment. Sometimes, a candidate is using two monitors or a separate, preferred IDE. Other times, however, they might be opening new tabs on another screen to secretly search for answers. Reviewers should combine the mouse exit count with other signals, like the number of tabs opened, to gain more insight into the candidate’s performance.
5. Use Facial Detection To Warn When Multiple People Are in Snapshots
It is common for developers to ask for help from their peers. When they get stuck, developers typically first look to Google and then their developer friends. However, they should do their interviews by themselves. When doing coding assessments independently, it is recommended to use the candidate’s webcam (with their consent) to snapshot any moments where multiple faces appear on camera during the assessment. Sometimes these other people are just innocently passing by, but other times it is clear that the candidate is getting help. Certainly, watching the full video of the candidate completing their assessment will help to identify this kind of fraud. But hiring managers can also use the number of snapshots flagged to identify the increased likelihood of fraud.
6. Discover if GitHub Is Empty or Created on the Same Day
Most real tech professionals have a public footprint that can be used to validate their identity. This is particularly true when assessing developers. With the advent of open source and social media, developers often leave a trail of code, commits and comments on their areas of expertise for the public to review. Importantly, many of these services now allow for single sign-on authentication. This offers a great opportunity to connect your candidate to their public profiles explicitly. Requiring candidates to authenticate themselves using LinkedIn, Facebook, GitHub, Bitbucket, or Gitlab can provide additional insight and will often provide photos for facial comparisons. Treat empty accounts with little public history as another red flag, as fraudulent developers are known to create a new GitHub account just to submit the interview. These accounts have a recently created date and no repos or activity, which may indicate they are trying to hide their identity.
7. Aggregate Multiple Submissions
Sometimes the same candidate applies to multiple jobs to take the same standardized test several times to ensure they can pass it. This is impossible to detect unless the testing provider delivers this level of analysis to flag any duplicate submissions. Of course, custom tests would obviate this, but part of the allure of interview platforms is that they provide a catalog of tests to help you hire for programming skills you may not already have in-house. If your company is reusing tests across many job openings, particularly for MSPs, you will be glad to have AI flag these candidates for further review.Â
How are you using AI to prevent fraud from damaging your business? Share with us on FacebookOpens a new window , TwitterOpens a new window , and LinkedInOpens a new window .
Image Source: Shutterstock