AI for Facilitating and Preventing Scams

Artificial intelligence is a double-edged sword: it can be used for fraud and scam detection or be the very reason of a scam or fraud case.  

A call or an e-mail you have received might sound or feel familiar, and anyone can be a target of this crime.

Apple co-founder Steve Wozniak confirms that artificial intelligence could make scams harder to identify. He believes that AI content should be clearly labeled, adding that text generators like ChatGPT could make bad actors seeking to trick people about who they appear more convincing. Indeed, regulatory gaps are making AI-enhanced content risky for different society clusters.  

Indeed, AI regulation is work-in-progress. Just like Tim Burton’s Beetlejuice character appearing after hearing its name thrice, European regulators are being summoned one by one after hearing about the “regulatory gap” just once. No doubt that soon there will be artificial intelligence rules overseeing the use of AI and the display of AI-enhanced products. Regardless, just as GDPR cannot prevent data leaks, AI regulation will likely not avert AI-driven scams, meaning we have to take the matter into our own hands. Moreover, as the technology develops further, leisure internet users (e.g., the society cluster that forwards the messages without fact-checking) and vulnerable clusters (kids, seniors, etc.) will have a hard time understanding what’s real and what’s not, especially during sensitive times like local or national elections, crises, and similar situations.

AI could be used to create fake data and attack targets continuously, like “The Masked Scammer” on Netflix, but more professional and systematic. 

A deepfake combines existing data (e.g., image) with separate data to create something new.

How can we protect ourselves (and others) from AI-based scams?

1. First rule of prevention is understanding vulnerabilities and potential scenarios, and this case is no different. Understanding possible attacks could be critical in these situations. For example, voice mimics and voice clone scams are widespread these days; however, the current infrastructure could be used to alter videos, images, documents, and websites, creating “deep fakes.” As technology evolves, there will be more ways of creating chaos, so staying on top of the trends is never a bad idea. 
2. Fake data is usually modeled by the existing data online. This means we should be more careful about submitting our personal data, including pictures, videos, voice recordings, etc. Not submitting data to unknown third-party platforms and keeping social media accounts in a private setting would set the barrier higher. Contact data submissions, including phone number or e-mail addresses, should also be monitored, as this data will direct the scammers to you or someone you know.
3. Differentiating scams is getting more complex; however, when there is a suspicious situation challenging the situation further would make sense.

• The first step would be to watch out for unknown, international, or weird caller IDs (or e-mail accounts). 
•  Then, during the call, watching out for unnatural and metallic sounds and weird pronunciations will keep you on your toes. The same would apply to scam texts, as spotting a typo or a word that doesn't belong in the text could reveal the scam.
•  If you suspect the caller, offer to hang up and call back if possible. You can also let unknown numbers go to voicemail as a screening measure.
• Following the detection of something unnatural, the logical step would be to question the caller further, requiring information that is not public, preferably only the real caller would know. If you have received suspicious documentation, don't hesitate to request additional proof and double-check.
• Also, keeping an eye on the communication patterns (call or e-mail time, frequency, etc.) could give you an idea about the authenticity of the call or the message.

Remember, if something doesn’t feel right, it probably isn’t and it is always worth a few seconds to double check a piece of information, news, website, or a call. 

Quoting Wozniak, "We can't stop the technology, but we can prepare people to be better educated" to spot fraud and malicious attempts to take personal information.

If you want to learn more, you can read German Federal Office for Information Security’s Deep-Fake Warning Page.