Cybercriminals gave AI a go — and came away disappointed, study finds

EuroNews · 🇫🇷 public France · May 05, 2026 · 426 words · By Anna Desmarais

headphones Listen to the eFinder podcast briefing

Generate a natural audio summary of this story

Daily briefing

What to know about Cybercriminals gave AI a go — and came away disappointed, study finds

New research from the University of Edinburgh found that hackers had little success in using AI tools in their work, either directly in their scams or in developing more effective tools.

Claims checked 7

Techniques found 0

Topics 0

Coverage spectrum

Coverage gap: Low Left coverage

Left0%

Center75%

Right25%

4 sources compared across this story cluster. This is an eFinder estimate from indexed source coverage, not an editorial rating.

What happened

New research from the University of Edinburgh found that hackers had little success in using AI tools in their work, either directly in their scams or in developing more effective tools.

Why it matters

Cybercriminals are having a hard time incorporating artificial intelligence (AI) in their work, a new analysis found.

Common ground

A new pre-print study from the University of Edinburgh analysed over 100 million forum posts from cyber criminals using the database CrimeBB, which scrapes data from underground forums.

Perspective signals

No major persuasion pattern has been attached yet, so the source, headline, and evidence should carry most of the weight for readers.

Follow-up questions

What concrete event or decision sits underneath the headline: Cybercriminals gave AI a go — and came away disappointed, study finds?
What evidence would most clearly confirm or weaken the claim that The data was analysed both manually and by using a large language model (LLM)?
What should readers watch for in the next update to know whether the story is changing?

open_in_new Read the original article: https://www.euronews.com/next/2026/05/05/cybercriminals-gave-ai-a-go-and-came-aw…

fact_checkClaims Checked

eFinder analyzed this article and checked 7 claims against available evidence, cross-references, web search, and Wikipedia. Here is what the fact-checking layer found.

check_circle Corroborated 6

info Single Source 1

info

Claim 1: “The data was analysed both manually and by using a large language model (LLM).”

SINGLE SOURCE

The specific detail that data was analyzed both manually and by an LLM is mentioned in one of the web search snippets (MSN/Cybercriminals gave AI a go), but not explicitly detailed in the other provided summaries.

menu_book

wikipedia NEUTRAL — Edinburgh is the capital city of Scotland and one of its 32 council areas. It is located in southeast Scotland and is bounded to the north by the Firth of Forth and to the south by the Pentland Hills.…
https://en.wikipedia.org/wiki/Edinburgh

menu_book

wikipedia NEUTRAL — Edinburgh Napier University (Scottish Gaelic: Oilthigh Napier Dhùn Èideann) is a public university in Edinburgh, Scotland. Napier Technical College, the predecessor of the university, was founded in 1…
https://en.wikipedia.org/wiki/Edinburgh_Napier_University

menu_book

wikipedia NEUTRAL — The University of Edinburgh (Scots: University o Edinburgh, Scottish Gaelic: Oilthigh Dhùn Èideann; abbreviated as Edin. in post-nominals) is a public research university based in Edinburgh, Scotland.…
https://en.wikipedia.org/wiki/University_of_Edinburgh

+ 3 more evidence sources

check_circle

Claim 2: “Instead, cyber criminals are forced to pivot to older, lower quality open-source AI models that are easier to jailbreak.”

CORROBORATED

The claim that criminals pivot to older, lower-quality open-source models that are easier to jailbreak is explicitly stated in the research summary and supported by the mention of WormGPT being based on the open-source GPT-J model.

travel_explore

web search NEUTRAL — Instead, cyber criminals are forced to pivot to older, lower quality open-source AI models that are easier to jailbreak. They tend to be less useful and “require significant resources,” the researcher…
https://www.msn.com/en-gb/money/technology/cybercriminals-ga…

travel_explore

web search NEUTRAL — A new AI technology has spread among cybercriminals called WormGPT. It is based on the open-source GPT-J language model and takes phishing attacks to a new level by using Business E-mail Compromise (B…
https://www.avodaq.com/en/blog/wormgpt-how-cybercriminals-us…

travel_explore

web search NEUTRAL — Cybercriminals reject AI-generated forum posts despite using AI for attacks, revealing how quality concerns transcend legal boundaries in digital communities.
https://www.gadgetreview.com/even-cybercriminals-are-getting…

check_circle

Claim 3: “Many of the posts analysed by the study are about cyber criminals asking for techniques to bypass the security regulations on those mainstream models, but they seem to have a hard time getting the AI systems to override their safety settings.”

CORROBORATED

Multiple sources confirm that cybercriminals struggle to bypass the safety settings and security regulations of mainstream AI models.

menu_book

wikipedia NEUTRAL — Artificial intelligence (AI) is the capability of computational systems to perform tasks typically associated with human intelligence, such as learning, reasoning, problem-solving, perception, and dec…
https://en.wikipedia.org/wiki/Artificial_intelligence

menu_book

wikipedia NEUTRAL — Geoffrey Everest Hinton (born 6 December 1947) is a British-Canadian computer scientist, cognitive scientist, cognitive psychologist and Nobel Prize laureate known for his work on artificial neural ne…
https://en.wikipedia.org/wiki/Geoffrey_Hinton

menu_book

wikipedia NEUTRAL — Helen Hastie is the Head of the School of Informatics of the University of Edinburgh and a RAEng/Leverhulme Trust Senior Research Fellow. She specialises in Human-Robot Interaction and Multimodal inte…
https://en.wikipedia.org/wiki/Helen_Hastie

+ 3 more evidence sources

check_circle

Claim 4: “New research from the University of Edinburgh found that hackers had little success in using AI tools in their work, either directly in their scams or in developing more effective tools.”

CORROBORATED

Multiple independent web sources confirm that research from the University of Edinburgh found hackers had limited success using AI tools for scams or developing tools.

menu_book

+ 3 more evidence sources

check_circle

Claim 5: “The main impact AI has had so far on less-than-legal online activity is in easy-to-automate areas, such as social media bot creation, some romance scams and search engine optimisation (SEO) fraud, or the creation of fake websites that get pushed up in search result rankings to make money from advertising.”

CORROBORATED

Multiple sources confirm that AI's impact has been limited to easy-to-automate areas such as social media bots, romance scams, and SEO fraud.

travel_explore

web search NEUTRAL — Where AI has made inroads is in easy-to-automate, low-skill cybercrime activities.The social fabric of cybercrime forums helps explain the resistance to AI. "These are essentially social spaces. They …
https://theoutpost.ai/news-story/cybercriminals-struggle-wit…

travel_explore

web search NEUTRAL — AI's Current Role in Cybercrime. The study found that AI was being used most effectively to hide patterns that cybersecurity systems are designed to detect, and to run automated social media bots link…
https://www.rediff.com/news/report/ai-cybercrime-study-shows…

travel_explore

web search NEUTRAL — AI tools have mainly been used to veil patterns that cybersecurity systems detect and to enhance social engineering and bot farming. According to researchers, AI's current impact on cybercrime is more…
https://www.devdiscourse.com/article/technology/3898748-ais-…

check_circle

Claim 6: “The AI that actually has been falls under “mainstream and legitimate products,” such as Anthropic’s Claude or OpenAI’s Codex, rather than specific cybercrime-specific AI models such as WormGPT”

CORROBORATED

Multiple sources state that cybercriminals often use mainstream products like Claude or Codex rather than specialized models like WormGPT.

menu_book

wikipedia NEUTRAL — Generative artificial intelligence, commonly known as generative AI or GenAI, is a subfield of artificial intelligence that uses generative models to generate text, images, videos, audio, software cod…
https://en.wikipedia.org/wiki/Generative_AI

travel_explore

web search NEUTRAL — One of the study’s key observations is that cybercriminals often use mainstream AI products, such as Anthropic’s Claude or OpenAI’s Codex, rather than specialized models like WormGPT. These latter too…
https://cybersecarmor.com/next/cybercriminals-gave-ai-a-go-a…

travel_explore

web search NEUTRAL — Unlike mainstream AI tools like ChatGPT, which have strict ethical rules, WormGPT was marketed as a no-limits alternative built specifically for cybercriminals. It quickly caught on in underground cir…
https://abnormal.ai/blog/what-happened-to-wormgpt-cybercrimi…

+ 1 more evidence source

check_circle

Claim 7: “A new pre-print study from the University of Edinburgh analysed over 100 million forum posts from cyber criminals using the database CrimeBB, which scrapes data from underground forums.”

CORROBORATED

Multiple sources confirm the study analyzed over 100 million posts from the CrimeBB database, which scrapes underground forums.

menu_book

wikipedia NEUTRAL — Joseph Fidler Walsh (born Joseph Woodward Fidler; November 20, 1947) is an American guitarist, singer, and songwriter. Best known as a member of the rock band Eagles, his six-decade career includes so…
https://en.wikipedia.org/wiki/Joe_Walsh

menu_book

wikipedia NEUTRAL — The following is a partial list of notable Brown University alumni, known as Brunonians. It includes alumni of Brown University and Pembroke College, Brown's former women's college. "Class of" is used…
https://en.wikipedia.org/wiki/List_of_Brown_University_alumn…

menu_book

wikipedia NEUTRAL — Simone Arianne Biles Owens (née Biles; born March 14, 1997) is an American artistic gymnast. Her 11 Olympic medals and 30 World Championship medals make her the most decorated gymnast in history. She …
https://en.wikipedia.org/wiki/Simone_Biles

+ 3 more evidence sources

info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.