STOCK IMAGE | Image by macrovector on Freepik

by Patricia B. Mirasol, Producer

The Department of Information and Communications Technology (DICT) advises netizens to report misleading social media posts, even as posts manufactured by deep fake technology proliferate in such platforms. 

Deepfake technology is a type of artificial intelligence that creates fake images, videos, or audio recordings through algorithms that stitch together such multimedia materials. 

“Let’s look out for one another and stay vigilant,” said Ramon S. Ang, chairman of San Miguel Corporation, in a February 17 post on his Facebook wall. 

According to Mr. Ang, a fraudulent sponsored post featuring his interview with broadcast journalist Anthony T. Taberna, Jr. has recently been making rounds.

“Scammers have used deepfake technology to alter our conversation, making it appear that I am endorsing an investment opportunity,” the post read. 

“It’s frustrating how easily these fake videos spread, deceiving people and taking their hard-earned money.” 

Facebook has not responded to BusinessWorld’s emailed query about deepfake content on its platform as of press time. 

 

Responsibility of social media platforms 

Even government officials are not exempt from deepfake-generated posts, according to Jeffrey Ian C. Dy, DICT undersecretary for infostructure management, cybersecurity, and upskilling. 

There’s a deepfake post of him selling modems, Mr. Dy told BusinessWorld in a February 19 text. 

The hearings at the Lower House by the tri-com, Mr. Dy said, “took in the task to hard questions about social media platforms’ responsibility to police posts.”  

During a February 18 investigation into online disinformation at the House of Representatives (HoR), Surigao del Norte Representative Robert Ace S. Barbers revealed that a committee is evaluating a proposal to mandate social media companies to obtain a franchise to operate, aiming to curb the spread of disinformation.

“There needs to be regulation,” Mr. Barbers said in the vernacular. “There should be a policy for our social media platforms and actors because they shouldn’t be allowed to use it to spread fake news or baseless statements.” 

The DICT, Mr. Dy said, has partnered with platforms like Meta, Google, and TikTok to promote verified accounts, or accounts that have been verified as authentic by the platforms. 

“We also have a working relationship with fact checkers to empower them,” he said in the same February 19 text. “Yung nga lang, these solutions usually tend to act on sensational deepfake posts.” 

 

Recognizing Deepfakes 

On a February 4 House hearing, cybersecurity analyst Jan Marcelo “Marco” S. Reyes of the DICTs Cybercrime Investigation and Coordinating Center (CICC) said the agency has a tool that can help law enforcement agencies detect deepfake videos. 

The CICC has acquired a tool called Aletheia Deepfake Agent to detect deepfake videos, Mr. Reyes told BusinessWorld in a February 20 Viber message.  

Once activated on a Windows computer, it automatically analyzes any video played on the screen—whether from Facebook, YouTube, or even Zoom. The software can flag potential deepfakes and notify the user within half a minute. 

“Right now, CICC has secured several licenses and is focusing on empowering online communities with large followings to help verify suspicious content,” Mr. Reyes said. 

“If someone comes across a possible deepfake, they can share it with these communities for quick validation,” he said. 

The CICC is looking into broader deployment, he added. 

The Massachusetts Institute of Technology’s media lab, via its Detect DeepFakes campaign, offers tips on how to better spot a deepfake:  

  • Pay attention to the face. High-end Deepfake manipulations are almost always facial transformations.  
  • Pay attention to the cheeks and forehead.  Deepfakes may be incongruent on some dimensions. 
  • Pay attention to the eyes and eyebrows. Deepfakes may fail to fully represent the natural physics of a scene.  
  • Pay attention to the glasses. Deepfakes may fail to fully represent the natural physics of lighting. 
  • Pay attention to the facial hair or lack thereof. Deepfakes may fail to make facial hair transformations fully natural. 
  • Pay attention to facial moles.  Does the mole look real?  
  • Pay attention to blinking. Does the person blink enough or too much?  
  • Pay attention to the lip movements. Some deepfakes are based on lip syncing.

“We can report [such posts] to Facebook and Facebook takes them down, but that can take weeks,” Mr. Dy said.  

“We can also report these cases to our hotline 1326. We collate these reports and act on them,” he added.

RELATED ARTICLESMORE FROM AUTHOR