ainewsblitz.com

Breaking

Study Finds Claude and GPT-4o Outperform Humans at Building 'Exploitable Trust' in Simulated Romance Scams

Study Finds Claude and GPT-4o Outperform Humans at Building 'Exploitable Trust' in Simulated Romance Scams
  • Security
  • AI Agents

In a first-of-its-kind week-long experiment, researchers found that large language models were more effective than human operators at cultivating the kind of emotional trust that underpins romance-baiting fraud, raising alarms about how easily such scams could be automated at scale. The study, published on arXiv, pitted AI chatbots built on Anthropic's Claude Sonnet 3.7 and OpenAI's GPT-4o against human operators in a blinded, seven-day text-messaging trial designed to mimic the early stages of so-called "pig butchering" scams.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.

$20
Read this article
$29/month
Unlimited — all 27,934 articles, the full archive, and comprehension quizzes
Save 72%
$98/year
≈ $8.17/month
Unlimited, billed once a year