Safe Words and Shared Secrets: The Low-Tech Shield Against AI Deepfakes
As AI-generated clones become indistinguishable from reality, experts urge a return to non-digital verification protocols.
The era of trusting your eyes and ears is over. As generative AI makes deepfakes indistinguishable from real human interactions, security experts are advising a return to 'low-tech' defenses to prevent identity theft and fraud.
To combat the rise of sophisticated cloning, experts now recommend the use of shared secrets or 'safe words' to verify identity during suspicious video or audio calls. This method provides a definitive verification step that does not rely on software. According to Deepak Gupta, CEO of GrackerAI, "Seeing and hearing someone is no longer proof they are real."
The Erosion of Digital Trust
This shift comes as generative AI allows attackers to create highly convincing audio and video clones of family members and corporate executives. Historically, deepfakes could be spotted by 'tells' such as unnatural blinking patterns or subtle audio glitches. However, these technical flaws are rapidly disappearing as AI models evolve, leaving traditional detection methods obsolete.
Why Software is Failing
Technical detection tools are currently struggling to keep pace with the speed of AI evolution. Most AI-detection software relies on probability—calculating the likelihood that a clip is synthetic—which creates a constant arms race between the generator and the detector. In contrast, low-tech security protocols, such as pre-arranged shared secrets, offer a more reliable defense because they rely on exclusive knowledge rather than algorithmic probability, as reported by ZDNET.
The Human Protocol
This transition shifts the burden of security from software to human protocol. If technical detection cannot keep up with AI generation, the only remaining reliable verification method is a non-digital secret known only to the parties involved. By implementing these protocols, organizations and families can create a scalable defense that does not require constant software updates to remain effective.
What to Watch
As corporate fraud and identity theft increase, the industry is watching to see if these manual protocols will be formalized into enterprise security standards. While shared secrets provide a current stopgap, the long-term challenge remains: establishing a baseline of trust in a digital environment where visual and auditory evidence can be fabricated in seconds.