Science & Tech · Cybersecurity · 12 hrs ago
Researchers develop real-time warning system for video-call deepfakes
Researchers at the Fraunhofer Institute for Secure Information Technology in Darmstadt, Germany, have developed a system that warns users about deepfakes during video calls.
It uses artificial intelligence to look for differences between real video and AI-generated images, and aims to detect fakes almost in real time.
That speed matters because a warning during a call could help someone avoid being tricked.
Criminals can use deepfakes to impersonate people and push employees into sending money or sharing information.
In a case in Hong Kong, scammers reportedly impersonated a company’s entire board in a video call and persuaded an employee to transfer $25 million to an account they controlled.
The system is still being tested internally.
The researchers plan to work with companies and video-call providers, including Microsoft Teams and Zoom, to prepare it for use with attention to usability and data protection.
Researchers at the Fraunhofer Institute for Secure Information Technology in Darmstadt have developed an AI-based system to warn users about deepfakes during video calls.
The system aims to distinguish real people from fakes in near real time so it can alert potential victims during a call.
The warning program is still in an internal testing phase.
The researchers plan to work with companies and video-call providers such as Microsoft Teams and Zoom on making it available.
The article cites a Hong Kong case in which criminals used a video call to impersonate a company’s board and induce an employee to transfer $25 million.
- Who
- A team led by Martin Steinebach at the Fraunhofer Institute for Secure Information Technology developed the system, with doctoral researcher Raphael Frick helping to develop its detection technology.
- What
- An AI-based warning system detects video-call deepfakes in near real time and aims to alert users during a call.
- When
- The system is in an internal testing phase; a launch date is not stated.
- Where
- The system was developed at the Fraunhofer Institute for Secure Information Technology in Darmstadt, Germany.
- Why
- To warn potential victims during video calls and help prevent deepfake-enabled fraud.
This story does not have two clearly opposing sides.
Real cases of fraud are now plentiful.
It opens up a completely new vector for fraud that is still difficult to deal with.
Steinebach’s team at the Fraunhofer Institute for Secure Information Technology developed the AI-based warning system.
The warning program is undergoing internal testing.
The team plans to work with companies and video-call providers such as Microsoft Teams and Zoom.
- Development site
- Fraunhofer Institute for Secure Information Technology, Darmstadt
- Detection speed
- Near real time
- Current status
- Internal testing
- Planned partners
- Companies and video-call providers, including Microsoft Teams and Zoom
- Hong Kong fraud case
- $25 million transferred to a foreign account after criminals impersonated a company’s board in a video call
Sources
Related news
New characters let lookalike domains bypass some browser checks
Hackers use Bing redirects in Google ads to promote fake Claude installers
Amazon Bedrock flaw exposed AI agent credentials and source code
AhnLab to Enable AI and Internet Access on Research Foundation Work Network
AI system finds 17 security flaws in tests of 50 mobile apps
US seizes two China-linked hacking tools used against critical infrastructure
Canadian cybersecurity negotiator arrested in ShinyHunters investigation
Anthropic offers free AI vulnerability scans to open-source projects
FBI arrests another suspect linked to ShinyHunters jobs-site breach
119 databases from Russian companies published online in first nine months of 2026
Study finds car apps share drivers’ data with advertising and analytics firms
