In today’s digital age, the line between humans and machines is becoming increasingly blurred. Artificial intelligence (AI) has advanced so much that it can now generate text, images and videos that are virtually indistinguishable from those created by humans.
This development poses a significant challenge to traditional verification methods, such as CAPTCHAs, which are no longer sufficient to guarantee that whoever interacts online is really a person.
A recent and compelling example of this problem is the case of Reddit, where researchers from the University of Zurich used AI bots to post more than 1,700 comments on the “Change My View” subreddit, deceiving both users and moderators.
This incident not only highlights the ability of AI to manipulate online discussions, but also highlights the urgent need to develop new verification methods that are resistant to these technologies.
AI and the obsolescence of Traditional Methods
Generative AI has crossed a critical threshold. Since it not only facilitates the creation of deepfakes and false content, but it also allows Malicious actors access platforms that were previously protected by simpler verification methods.
In the case of Reddit, AI bots not only posted comments, but also adopted fictitious personas, such as rape victims or activists against the Black Lives Matter movement, to influence users’ opinions.
This type of manipulation not only distorts discussions, but alson erodes trust in the authenticity of online content.
According to a report by iProov, face swap attacks (face swaps) increased by 704% between the first and second half of 2023, highlighting the growing sophistication of AI-powered attacks.
In addition, the proliferation of these technologies puts at risk the security of financial transactions, access to government services and the integrity of social networks, since bots can be used to commit fraud, spread disinformation or even for more sophisticated attacks such as phishing.
AI-resistant verification methods
Given this scenario, companies and developers are working on implementing more advanced and secure verification methods. Below are some of the most promising strategies:
Biometric authentication
Biometrics uses unique characteristics of the individual, such as facial recognition, iris scanning or fingerprints, to verify identity.
To prevent the use of deepfakes, life detection is incorporated, which verifies whether the person is physically present at the time of verification. This may include asking the user to perform certain actions, such as moving their head or blinking, to confirm that it is not a static image or pre-recorded video.
Practical example and not a manipulated image.
Behavioral biometrics
This method analyzes behavior patterns of the user, such as typing speed, mouse movements, or walking (on mobile devices).
Because these patterns are inherently inconsistent and difficult to replicate, they are an effective barrier against AI. AI strives to emulate natural variations in human behavior, making this method particularly robust.
Practical example: FTX Identity uses behavioral biometrics to detect bots, analyzing interaction patterns that are unique to each user.
Multimodal Verification
This combines multiple verification methods, such as biometrics, document analysis, and behavioral biometrics, to increase security.
Requires the AI to overcome several obstacles simultaneously, making it much more difficult to outwit. For example, a system might require the user to present an official document, complete a biometric verification, and pass a behavioral test.
Practical example: Reddit plans to work with third-party services such as Persona, Alloy, Stripe Identity, Plaid and Footprint, which integrate multimodal verification to ensure the “humanity” of users.
Blockchain for identity verification
Blockchain technology enables the creation of decentralized digital identities, where users control their own identity information.
Verifiable credentials, stored in a digital wallet, can be selectively shared with third parties, ensuring that only necessary information is revealed.
Since the blockchain is immutable and transparent, any forgery attempt would be immediately detectable. This ensures that digital identities are trusted and tamper-resistant.
Practical example: Telefonica Tech is exploring the use of blockchain for self-sovereign identity, allowing users to verify their identity without revealing sensitive data (Blockchain Identity).
AI generated content detection
Specialized tools can detect whether a text, image, audio or video has been generated by AI, using machine learning algorithms to identify patterns and characteristics typical of automatic generation.
These tools help filter out unwanted content, ensuring online interactions are authentic.
Practical example: Tools like Google SynthID can identify text generated by models such as ChatGPT, Gemini or Claude with high precision.
The future of online verification
Emerging technologies such as eye scanning or neural biometrics could become standards in the near future.
For example, the Orb eye scanning device from Tools for Humanity (a startup by Sam Altman) promises more robust humanity verification by requiring unique biometric data and physical presence.
However, these advances also pose ethical and privacy challenges, as they involve the collection and storage of sensitive biometric data.
In addition, educating users about the risks of AI and how to protect themselves will be essential to maintaining a safe and reliable digital ecosystem.
Only one thing is certain, the battle between humans and AI bots in the field of verification is just the beginning of a technological arms race.
Ultimately, the goal is to create a digital environment where authenticity and trust are the norm, not the exception, ensuring that online interactions remain a human domain, not a machine-generated simulation.
This post is also available in: