![\"\"](\"https:\/\/www.newsbeep.com\/il\/wp-content\/uploads\/2025\/11\/110325_MM_Sohni_Rais_014.jpg\")
11\/03\/25 \u2013 BOSTON, MA \u2013 Sohni Rais, a masters student studying information systems, works on a project inside the ISEC building on Nov. 3, 2025. Photo by Matthew Modoono\/Northeastern University<\/p>\nOne of the things that AI doesn\u2019t have that humans have in abundance is fingerprints.<\/p>\n
Researchers at Northeastern University used the unique fingerprints of human writing \u2014 word choice variety, complex sentences and inconsistent punctuation \u2014 to develop a tool to sniff out AI-generated text.<\/p>\n
\u201cJust like how everyone has a distinct way of speaking, we all have patterns in how we write,\u201d says Sohni Rais, a graduate student in information systems at Northeastern and a researcher on the project. In order to distinguish between human writing and AI text, she says, \u201cwe just need to spot the telltale patterns in writing style.\u201d<\/p>\n
11\/03\/25 \u2013 BOSTON, MA \u2013 Sohni Rais, a masters student studying information systems, works on a project inside the ISEC building on Nov. 3, 2025. Photo by Matthew Modoono\/Northeastern University<\/p>\n
![\"\"](\"https:\/\/www.newsbeep.com\/il\/wp-content\/uploads\/2025\/11\/110325_MM_Sohni_Rais_007.jpg\")
11\/03\/25 \u2013 BOSTON, MA \u2013 Sohni Rais, a masters student studying information systems, works on a project inside the ISEC building on Nov. 3, 2025. Photo by Matthew Modoono\/Northeastern University
\nWhile AI agents tend to write at a very consistent reading level, humans naturally vary.\u00a0\u201cWe might write simply when texting a friend but more formally in an email to our boss,\u201d says information systems graduate student Sohni Rais. Photos by Matthew Modoono\/Northeastern University<\/p>\n
AI text detection typically requires substantial computer power in the form of neural network transformers, says Rais, because these approaches analyze every letter, word and phrase in extreme detail. But this level of analysis isn\u2019t necessary to distinguish between human and AI-generated text, Northeastern researchers say. In fact, the technically \u201clightweight\u201d tool Rais helped develop can run on a regular laptop and is 97 percent accurate.<\/p>\n
\u201cWe are not the first in the world who develop detectors,\u201d says Sergey Aityan, teaching professor in Northeastern\u2019s Multidisciplinary Graduate Engineering Program on the Oakland campus. \u201cBut our solution requires between 20 and 100 times less computer power to do the same job.\u201d<\/p>\n
Existing AI-text detecting services, including ZeroGPT, Originality and AI Detector, train large language models to analyze each word. Text entered into these tools is analyzed by proprietary algorithms trained with large datasets powered by transformers.<\/p>\n
The lightweight tool can be trained by the user and live on their laptop, offering security and customizing advantages.<\/p>\n
\u201cEither you don\u2019t want your secret information to go somewhere beyond your laptop,\u201d says Aityan, \u201cor you are a professor and you want to catch your students cheating, so you train your own dataset based on specific texts.\u201d\u00a0<\/p>\n
Instead of using transformers, the lightweight approach uses 68 unique stylometric features \u2014 or \u201cwriting fingerprints,\u201d as Rais calls them \u2014 that make each person\u2019s writing unique. These features include sentence complexity.<\/p>\n
While AI agents tend to write at a very consistent reading level, humans naturally vary, she says.\u00a0<\/p>\n
\u201cWe might write simply when texting a friend but more formally in an email to our boss,\u201d Rais says.<\/p>\n
The tool also looks at word variety, which humans naturally mix up.\u00a0<\/p>\n
\u201cWe might say \u2018happy,\u2019 then \u2018glad,\u2019 then \u2018pleased,\u2019\u201d Rais says. \u201cAI often gets stuck using the same words repeatedly despite knowing many synonyms.\u201d<\/p>\n
It also looks at how far apart related words are in a sentence, she says. For instance, in \u201cthe cat that I saw yesterday was orange,\u201d the subject (cat) and the verb (was) are separated by five words. Sentences generated by AI, Rais says, maintain consistent distances of two or three words between subjects and verbs.<\/p>\n
Instead of looking at every single word, the lightweight approach looks for the most relevant clues.<\/p>\n
\u201cIt\u2019s like taking a person\u2019s vital signs at the doctor,\u201d she says. \u201cInstead of running every possible test, we measure key indicators like temperature, blood pressure and heart rate that tell us what we need to know.\u201d<\/p>\n
The work to develop ways of detecting AI-generated text isn\u2019t over, says Aityan. It is the nature of AI-based systems, however, to learn and improve, he says. As soon as people developed the technology to generate AI text, he says, the technology to detect it followed. And shortly after that, he says, came so-called humanization algorithms to make AI-generated text sound more natural.<\/p>\n
\u201cIt\u2019s an ongoing battle,\u201d he says.<\/p>\n","protected":false},"excerpt":{"rendered":"One of the things that AI doesn\u2019t have that humans have in abundance is fingerprints. Researchers at Northeastern…\n","protected":false},"author":2,"featured_media":142448,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5],"tags":[345,114,85,46,41723,1360,4277,39748],"class_list":["post-142447","post","type-post","status-publish","format-standard","has-post-thumbnail","category-business","tag-ai","tag-business","tag-il","tag-israel","tag-oakland","tag-research","tag-text","tag-writing"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/142447","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/comments?post=142447"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/posts\/142447\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media\/142448"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/media?parent=142447"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/categories?post=142447"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/il\/wp-json\/wp\/v2\/tags?post=142447"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}