Article Read Time
Artificial intelligence has exploded in the past couple of years. The emergence of advanced chatbots has benefited us all, but complex deepfakes have also given cybercriminals a faster, more effective way to perpetrate their crimes.
AI can create new malware code in minutes, compared to the hours a threat actor would take to make one manually. They can take audio and video files of real people (even YOU!) and generate fake clips of them posting, speaking, and performing various actions.
The latter is a cyberattack known as a deepfake. It is projected that 8 million deepfakes will spread this year.
Many high-profile deepfake cases demonstrate the severity of artificial intelligence-driven cyberattacks. Here are some examples you may have heard about.
- International pop star Taylor Swift was among many celebrities who faced legal battles following the widespread spread of deepfake videos.
- Popular rapper Megan Thee Stallion sued a YouTube blogger in late 2024 for circulating inappropriate deepfake videos of her.
- The principal of a Maryland high school was the target of a deepfake audio recording that seemed to depict him making racist comments.
Deepfakes can be the lead-up to a phishing scam, especially in the business world; a company in Hong Kong found this out the hard way when an employee sent $25M to a threat actor who convinced them that they were on a call with multiple colleagues and the Chief Financial Officer. Everyone on the video chat was a deepfake.
How to Recognize Deepfakes
Detecting deepfakes can be challenging, especially as artificial intelligence continues to advance — and AI accelerates all the time. There are still a few tells, however, that can clue you in that something you’re listening to, or watching, is actually a deepfake.
- Look for inconsistencies in facial expressions, blinking patterns, or lip-syncing. Deepfakes often struggle with natural eye movements and mouth synchronization.
- Pay attention to the skin texture and lighting. Deepfakes sometimes have an unnatural smoothness or inconsistent lighting across the face and body.
- Look for inconsistencies in details like hair, teeth, and jewelry. Deepfakes might struggle to render complex elements accurately.
- Check for mismatched lighting and shadows. Deepfakes often have lighting that doesn’t align naturally with the rest of the image.
- Watch for visual glitches, such as blurring or edge distortion, especially during fast movements.
- Examine reflections, even in small details like someone’s eyes. Inconsistent or unnatural reflections can be a sign of manipulation.
- Audio deepfakes might sound robotic or lack the natural tone and pitch variations of human speech.
- Listen for unnatural background noise or sudden changes in the audio environment.
Always verify the context and source of an audio or video file. Something that seems unusual or out of character might be a deepfake.
If you want to guarantee that something like that could never happen to you, toggle your privacy settings on social media so that only trusted parties can see your posts. Remember that nothing is ever truly erased from the Internet, so any videos, pictures, and audio clips posted can be recovered with the right tools and used to create a deepfake. Still, it’s helpful to periodically review your social media profiles and remove any posts you don’t want on your accounts.
Commonwealth Sentinel will help you face your organization’s growing cyber security threats. We can evaluate your existing IT security and work with your team to protect your data and assets.
At Commonwealth Sentinel, we are focused on cyber security so that you can focus on other things. Contact us today or sign up for a free consultation.