5 Hrs
Source:
وكالة صدى نيوز
Source: وكالة صدى نيوز
5 Hrsوكالة صدى نيوز
Source: وكالة صدى نيوز
5 HrsReady to play
Ready to play
A U.S. study found that generative AI tools can analyze public posts and images on social media and craft personalized phishing messages based on people’s interests and experiences—without hacking their accounts. Researchers tested about 18,000 AI-generated messages based on the accounts of 200 Instagram users. In an experiment involving 70 participants, the automated messages seemed less suspicious, scoring 2.75 out of five compared with 4.44 for genuine scam messages. Security tools using their default settings detected between 0% and 19% of the malicious prompts; for one tool, the rate rose to 98.7% after security instructions were added. The researchers stressed that these results do not measure how successful the scams would be in practice, but highlight the importance of detecting fraudulent intent before messages are generated.
Notice: This Is an AI-Generated Summary
Comments (0)