Black Friday and Cyber Monday have kicked off the holiday shopping season, but the pervasive presence of AI has consumers concerned about potential seasonal scams.
According to a recent McAfee survey, 88% of U.S. consumers believe that hackers will leverage AI to develop convincing online scams during the holiday period. Additionally, 57% anticipate that scam emails and messages will become more credible, while 31% are worried about distinguishing between legitimate messages from retailers or delivery services. The survey, which collected responses from 7,100 adults across the U.S., Australia, India, U.K., France, Germany, and Japan in September, highlights these apprehensions.
The use of AI in crafting sophisticated fraudulent emails and communications has gained traction among scammers, as explained by McAfee’s Chief Technology Officer, Steve Grobman. Grobman emphasized the risks faced by digitally active consumers who may be shopping on unfamiliar websites during the holiday rush. He pointed out that attractive deals could potentially compromise individuals’ cautious instincts.
In response to these concerns, McAfee has developed various tools throughout the year to detect and prevent scams effectively. These tools include mechanisms to alert users about malicious links in messages and AI models designed to identify suspicious texts, such as requests for sensitive information like social security numbers during online transactions.
Grobman highlighted the evolving landscape of AI-driven fraud and the challenges posed by increasingly personalized and targeted scams that exploit human emotions and urgency. McAfee is actively working on technologies, including bots, to assist users in verifying the authenticity of messages and identifying potential scams.
The interview with Digiday delved into the broader implications of combating AI-generated fraud, emphasizing the need for continuous vigilance and adaptive strategies. Grobman stressed the importance of educating the public about the nuances of fraud detection and the inherent uncertainties in identifying malicious intent in online interactions.
As companies and individuals navigate the complex terrain of AI-driven fraud, Grobman underscored the need for proactive measures, such as starting interactions from official websites to mitigate risks associated with fraudulent activities like fake customer support schemes.
Moreover, the discussion touched upon the evolving landscape of open-source AI technologies and the challenges posed by the misuse of large language models by cybercriminals. Grobman highlighted the accessibility of data sets for training AI models, emphasizing the asymmetric advantage this provides to malicious actors.
In conclusion, Grobman emphasized the ongoing efforts to address the limitations and challenges associated with large language models and generative AI technologies. While acknowledging the power of AI, he underscored the imperative of recognizing and addressing the inherent complexities and vulnerabilities in AI-driven systems.
