It turns out that 'a fictitious face created by AI' is used in a company's marketing strategy



In recent years, it has become possible for AI to generate 'faces that are indistinguishable from real humans,' and AI-made facial photographs are increasingly being used for

political purposes . Meanwhile, it turned out that there are cases where the fictitious face created by AI is used even in the account that conducts marketing activities of the company on LinkedIn, a business SNS.

The latest marketing tactic on LinkedIn: AI-generated faces: NPR
https://www.npr.org/2022/03/27/1088140809/fake-linkedin-profiles

1,000-plus AI-generated LinkedIn faces discovered in probe • The Register
https://www.theregister.com/2022/03/28/ai_fake_linkedin_faces/

One day, Renee DiResta , who belongs to the Stanford Internet Observatory , a cyber research institute at Stanford University, received a message from a person named 'Keenan Ramsey' on LinkedIn. Ramsey is a growth specialist at RingCentral , a company that provides cloud services for enterprises. Is it? ' These messages are common in corporate marketing activities on LinkedIn, and DiResta wasn't interested in them either.

However, DiResta said that he remembered being caught in Ramsey's profile picture and observed it. Then, although it seems to be a typical face photo at first glance, AI such as 'I only wear one earring', 'Hair disappears in the middle or suddenly appears', 'Both eyes are just in the middle of the image', etc. He noticed that it fits perfectly with the features of the facial photo that he generated.

Included in the tweet below is Ramsey's profile picture. Shannon Bond , editor of NPR in overseas media, said that this photo is '1: the background is blurred' '2: both eyes are exactly in the middle of the image' '3: the hair blends into the background' It is pointed out that it has features that apply to AI-made faces, such as 'It disappears in the middle or suddenly appears' and '4: There is an earring on only one ear'.



DiResta was a researcher of Russian disinformation campaigns and conspiracy theorists' anti-vaccine campaigns, and was accustomed to seeing AI-made facial photos, so he was able to realize that Ramsey's profile photo was fake. However, recent research has shown that 'it is difficult to distinguish an AI face from a real face, but rather I feel that the AI face is more reliable.' Hany Farid, a digital media analysis expert at the University of California, Berkeley, asked the average person on the Internet,'Is this a real person or is it synthetically generated?' If so, their answer is essentially the same as accidental. ' The reason why AI-made faces are more trusted than actual faces is that they have the 'most average characteristics' that exist in human facial photographs, and it is easy to feel that they are 'faces that I saw somewhere.' I'm guessing it might be.

AI-generated faces are indistinguishable from real faces and are more reliable than real faces-GIGAZINE



In a subsequent survey, RingCentral didn't have an employee named 'Keenan Ramsey,' and there was no record in a company called Language I / O, where he was previously employed, and he earned a bachelor's degree in business administration. There was no fact that I was also enrolled at New York University. In other words, Ramsey was a completely fictional character, including not only LinkedIn's face photo, but also other profiles.

DiResta initially suspected that Ramsey's account was being used for phishing scams, attempting to steal sensitive information via some sort of link. However, after that, DiResta received a message from an account with a profile picture made by AI that claims to be a RingCentral employee, and also sees Ramsey's message from a person who seems to be a real RingCentral employee. I heard that you received the email you received.

From this point, it turned out that Ramsey's account was actually associated with RingCentral and was responsible for RingCentral's marketing efforts. The marketing method in which an account like Ramsey approaches a potential customer and, if the customer is interested, the employees in the company actually take over the interaction is called ' lead generation '. In the past, it was often done face-to-face, but due to the development of SNS and the influence of pandemics, the place of lead generation is shifting to online in recent years.



Intrigued, DiResta and his colleague postdoctoral fellow Josh Goldstein began researching profile pictures on LinkedIn. As a result, more than 1000 accounts with profile images that seemed to be generated by AI were confirmed, and those accounts could not find any trace of existence other than LinkedIn. These accounts had titles such as 'Business Development Manager', 'Sales Development Executive', 'Growth Manager', and 'Demand Creation Specialist', but to companies claiming to have been enrolled in the past and universities claiming to have obtained a degree. Did not have a record of that person.

RingCentral wasn't the only company claiming that an account with an AI-made face photo was 'working,' and more than 70 companies actually owned fictitious accounts. In response to NPR's inquiries, some companies 'hired an outside marketer' but did not allow the use of AI-made facial photos, and NPR's profile was made by AI. He was surprised to find out that his face photo was used.

Heather Hinton, RingCentral's chief information security officer, claims he is unaware that a fictitious account has been created with an AI-made face photo and has not endorsed the practice. 'This isn't the way we do business.' 'In response to this, we realize once again that technology is changing at a speed that even we, who are watching technology, can't keep up. We must be more vigilant about what we are trying to do and what vendors are trying to do on our behalf. ' RingCentral will review lead generation practices and take concrete steps to educate its employees.

Also, at Bob's Containers, a startup that reuses shipping containers at home and businesses, accounts with AI-made profile images, like RingCentral, were doing lead generation. CEO Bob Balderas also knew that he had hired a company called airSales to boost his business, and that airSales had created an account on LinkedIn called 'Bob's Containers Business Representative.' He thought that the account actually belonged to the people who work at airSales. Balderas said he had terminated his contract with airSales because he felt uncomfortable using AI-made faces in his business.

Jeremy Camilloni, CEO of airSales, employs an independent contractor in the provision of marketing services, which may create a LinkedIn profile 'at its sole discretion', but airSales may request it directly. Claims not to be involved. 'As far as I know, there are no specific rules regarding the use of profile pictures or avatars in LinkedIn, which is common for LinkedIn technical users,' Camilloni told NPR, but LinkedIn's

professional community policy . 'Don't create a fake profile or fake information about yourself.' 'Don't use images of others or images that don't show your appearance in your profile picture.' increase.



DiResta and colleagues have identified more than 10 companies on LinkedIn that offer marketing services using AI-made faces, and NPR has tried to contact these companies. Renova Digital, an advertising company that responded to the call, offered unlimited message transmission with a 'fully branded avatar profile' for a monthly plan of $ 1,300. This plan was removed after contacting NPR, and Renova Digital founder Philip Foti said in an email to NPR, 'We have a campaign with an AI face. We decided that it wasn't in line with our values and that it wasn't in the interests of marketing. '

After contacting DiResta and others, LinkedIn said it investigated accounts that violated rules such as fake profiles and tampering with information, and removed those that were found to be in violation. 'Our policy makes it clear that all LinkedIn profiles must represent a real person,' said Leonna Spilman, a LinkedIn spokeswoman. He claimed that he was working on improving the model so that he could detect the profile to use.

LinkedIn has also removed pages from LIA in India, which allegedly provided profiles with AI-made facial photos, and a San Francisco-based company called Vendisys. Information has already been deleted from LIA's official website, but in the past, a service was offered that allows you to select 'ready-to-use AI-generated avatars' from hundreds of candidates for $ 300 per month.



Using an AI face for marketing may be relatively harmless, but Farid eventually said, 'It's so high quality that it's hard to identify with the naked eye, with a full-fledged voice that imitates a specific person. He is concerned about the appearance of 'deep fake videos'. It has already been reported that 'a fake video of Ukrainian President Zelensky calling for surrender' has already appeared in connection with Russia's invasion of Ukraine.

'Fake footage of Ukrainian President Zelensky calling on people to surrender' has been released and Facebook has been deleted --GIGAZINE



Related Posts:

in Software, Posted by log1h_ik