Deepfake detection firms buy, build and buddy-up to combat growing threat

As new deepfake tools become increasingly accessible, there is a growing risk of undermining trust and security. Even cyber hackers with minimal skills can use these tools to carry out scams.
With the proliferation of AI-generated audio and video content on social media, companies are now seeking to incorporate deepfake detection models into their security infrastructure. This is especially important as synthetic content can even bypass biometric facial verification.
Reality Defender has engaged in collaborative efforts with Respeecher and ElevenLabs with the aim of enhancing its detection platform. Trend Micro has implemented upgrades to its Trend Vision One platform in order to address the GenAI deepfake content. Finally, Hiya has acquired Loccus.ai with the intention of bolstering its voice security platform.
Reality Defender partners to combat audio deepfakes
Reality Defender has partnered with Respeecher to enhance audio deepfake detection models. Through this collaboration, Reality Defender’s clients will have access to Respeecher’s advanced generative AI capabilities integrated with Reality Defender’s detection platform.
The jointly developed system will address the challenge of AI-generated content by identifying and thwarting audio threats. According to the company, Respeecher’s audio processing models produce synthetic speech that will enhance the effectiveness of the detection platform.
“Our proprietary audio processing models will significantly enhance the accuracy and effectiveness of Reality Defender’s detection platform, providing an unparalleled level of protection against AI-generated threats,” says Alex Serdiuk, chief executive officer and co-founder of Respeecher.
To further strengthen the audio deepfake detection models, Reality Defender partnered with ElevenLabs. In addition to safeguarding against audio deepfakes, the two companies are collaboratively working to reinforce the security of AI tools.
ElevenLabs is contributing its proprietary models and methodologies, which, when combined with Reality Defender’s multi-model detection approach, enhance the dependability of our detection models, the company reports.
“Collaborating with fellow technology companies, academia, and policymakers to develop reliable methods for AI content detection is a key pillar of our safety efforts,” says Aleksandra Pedraszewska, head of AI Safety at ElevenLabs.
Earlier this year, Reality Defender expanded its capabilities by adding support for Spanish and Portuguese audio files as part of its effort to develop a multilingual deepfake detection model. Additionally, the company enhanced its video deepfake detection models with the introduction of a new transformer-based video model called “Guided,” which emphasizes facial representations based on real samples.
Trend Micro upgrades Trend Vision One to prevent deepfake scams
Trend Micro has introduced improvements to its Trend Vision One platform to safeguard individual consumers from AI-powered impersonations and scams. The Deepfake Inspector is currently accessible to consumers and will soon be integrated into the Trend Vision One platform.
According to the company, the deepfake detection system goes beyond traditional image noise analysis and color detection by incorporating user behavioral analysis. Once a deepfake is detected, the platform notifies enterprise security teams to proactively counter future attacks.
“With our newly announced cutting-edge capabilities within the Trend Vision One platform, we are at the forefront of combating these threats for our enterprise customers, to detect deepfakes and other forms of AI fraud,” says Mick McCluney, ANZ Field chief technology officer, Trend Micro ANZ.
The latest features in the Trend Vision One platform enable centralized management of employees’ GenAI access and usage. The platform also provides tools to inspect prompts in order to prevent data leaks and malicious injections.
These updates offer real-time verification in live video conversations, alerting users to potential deception. A Trend Micro survey revealed that approximately 36 percent of consumers experienced scam attempts involving deepfakes.
Hiya acquires Loccus.ai for AI voice detection system
Hiya, a provider of voice security services, has acquired Loccus.ai, a company specializing in deepfake voice detection systems. This move is aimed at enhancing the Hiya voice security platform through the incorporation of AI-generated voice detection capabilities.
Following the acquisition, the product offering will be rebranded as Hiya AI voice detection, which will complement the existing Hiya adaptive AI fraud prevention system. The company states that the AI voice detection software is capable of identifying AI-generated voices with 1.5 seconds of audio.
Moreover, for broader deployment, the deepfake detection system can be integrated into personal communications apps, news and social media content, video conferencing, and contact center platforms, the company says.
“Joining Hiya allows us to help more customers leverage Loccus.ai voice detection with the industry-leading voice security offering, as we scale to 500+ million global users across the Hiya network,” says Manel Terraz, chief executive officer and founder of Loccus.ai.
Article Topics
biometric liveness detection | deepfake detection | deepfakes | Hiya | Reality Defender | synthetic voice | Trend Micro
Comments