Deepfake detection

What Is Knowledge Distillation? Knowledge distillation (KD) is a process of transferring knowledge from a more sizable neural model to a smaller one. A more robust model is used to “teach” a simpler student neural network. This is done to cut the computational and deployment costs whenever a model needs to be adapted for a

Can People Tell If a Text Is Created by AI? Human ability to detect GenAI content has been a topic of discussion since at least 2017, when the first generative models capable of producing realistic images became publicly available.  With the rise of other advanced generative models — GPT, Bark, stable diffusion —  identifying synthetic

What is an LLM Watermark Spoofing Attack? Contrary to the removal attack, watermark spoofing aims at producing some type of harmful content — such as dangerous advice, toxic statement, or biased info — that is falsely validated with a legitimate watermark of a certain Large Language Model (LLM). In turn, such a spoofing attack can

What Is LLM Watermark Erasing (Removing) Attack? Watermark erasing attack is an unauthorised process of removing metadata inserted in some type of media: video, image, text, and so on. Large Language Models (LLMs) and their output are also watermarked. The attack is typically orchestrated without tempering with the encryption key applied to protect the media.

Why is AI-Generated Content Detection Important? Content produced by Large Language Models (LLMs) lacks human expertise and understanding, which often makes it unreliable and even dangerous. A GenAI model, ChatGPT is known to provide false and harmful medical advice, discriminative statements, incorrect legal information, and so on.  Besides, it can be used for making fake

What Is Text Watermarking? Text watermarking is a technology, which allows copyright protection with the help of special metadata inserted into a copy of a text. This metadata contains details on authorship, owner status, publication date, and other info that can be extracted to avoid unauthorised distribution, counterfeiting, intellectual theft, and so on. While the

The problem of bias — related to age, ethnicity, or race — has been discussed in relation to AI since robust facial recognition models have become widespread. Currently, the problem can be also observed in the way the text detectors function — they are suspected to be biased towards non-native English authors.  In turn, this

Can AI-Generated Text Be Detected by Plagiarism Checkers? Plagiarism concerns in regard to GenAI have been raised repeatedly, especially since models capable of synthesizing media have become publicly available. Currently, AI-written texts aren’t acknowledged as intellectual property or original writing as they draw ideas, narratives, and stylistics from other people’s works. While the plagiarism and

What Are the Main Methods of Natural Language Watermarking (NLW)? NLW is a technique that protects a writing with a special signal that is invisible to a regular reader. This signal usually contains metadata pertaining to the text: date of release, authorship, genre, copyright info, and so on. Among the main NLW methods are the

What Is Natural Language Watermarking (NLW)? Natural Language Watermarking (NLW) is a technique, which helps labeling a text with a special signature invisible to a human eye. In essence, this measure helps prevent plagiarism and data leakage (“traitor tracing”), as well as allows tracking down tampered information and even performing text anti-spoofing. However, NLW is

What Are Adversarial Attacks in Natural Language Processing (NLP)? Adversarial attacks in NLP are a malicious practice of altering text with slight perturbations which can lead to a poor or inadequate performance of a text-based AI system. These perturbations include deliberate misspelling, rephrasing and synonym usage, insertion of homographs and homonyms, back translation, and so

Is It Possible to Detect AI-Generated Text — And Why Is It Necessary? Opinions vary on whether detection solutions can effectively spot AI texts. A skeptical view would take the position that at some point the difference between human writing and AI-made content will disappear as neural models progress at a lighting speed.  However, there

What Are Data Poisoning Attacks? With the recent boom of AI and Large Language Model (LLM) usage among the public, apprehensions have increased as to whether these tools could be used by bad actors for destructive purposes. These concerns have merit, as there are many ways these models could be compromised.  One such malicious technique

Deepfake Forensics: Definition and Overview Deepfake forensics is a subset of digital forensics — a group of techniques and approaches to detect falsified content, especially content used for legal purposes. These methods have been in development since at least 2005, when researcher Hany Farid suggested using color filter array artifacts left by digital cameras to

Understanding Digital Watermarking Digital watermarking is a technique to protect authenticity of digital media: pictures, video, audio, as well as mixed media. The concept’s main idea is to insert a specific hidden signature into the media — such as a news footage — that is invisible to the viewer, yet identifiable by a system. The

Definition and Problem Overview A social media stream refers to a feed or wall of content gathered from various digital platforms and then streamed through a specific online channel. Such a channel can be a digital signage, social media aggregator, or a similar outlet: some examples include YouTube Live, Hootsuite, and so on. A physical

Example from https://thispersondoesnotexist.com: Although the image appears ultra-realistic, there are discernible artifacts suggesting its deepfake nature. The reflections in the pupils are not identical, and the left and right lenses of the glasses are not symmetrical. History of the Term Deepfake (derived from ‘deep learning’ and ‘fake’) is a falsified synthetic media — video, photo,

Definition and Problem Overview As new applications for AI continue to expand, so does the concern over its potentially questionable — or even dangerous — applications. AI Ethics is a new field in the area of Artificial Intelligence which addresses these moral issues raised by AI innovations. While AI adds to the toolkit of malicious

Definition and Problem Overview Even though advanced Artificial Intelligence (AI) has been known to the broader public since at least 2017, the need for its legal control has been realized just recently, leading to a concept of Ethical AI. It asserts that successful AI legal regulation has five elements: AI has been acknowledged as a

Definition and Problem Overview The potentially disastrous consequences of Artificial Intelligence (AI) has been a topic of discussion since at least the late 19th century – that was when R.C. Reade’s novel, The Wreck of the World, captured the public’s interest by describing a rebellion of self-aware machines against humanity. Similar fears were addressed in

The first technology similar to facial deepfakes appeared in 1997 when the Video Rewrite tool was presented. It was based on automatic phoneme labeling that allowed matching an already existing footage to a new soundtrack. The tool was successfully applied to alter a few bits

Deepfake media have shown an alarming increase in recent times. According to expert reports, the amount of that type of media, including facial deepfakes, doubles every six months, as the tools and means to produce such fabricated media are becoming greatly available to the

Deepfake detection technology allows prompt recognition of a piece of fabricated media with high accuracy based on liveness signals. Distinguishing fake footage, photo or audio from authentic media is also necessary, as it helps to tackle the so-called liar’s dividend

Cheapfake — ( a combination of cheap and deepfake) — are a class of fake media that is easy and quick to produce, especially for amateur users. Compared to the traditional deepfakes, cheapfakes do not require highly specific skills such as coding, manual neural network

Deepfake dataset is a collection of artificially synthesized media, which can include photo, video, and audio materials designed in accordance with the deepfake standardization. Their origin was urged by the proliferation of deepfake media that represents a steadily growing social threat

Deepfake detection is a capability of recognizing falsified media and distinguishing it from bona fide visual or auditory data. Deepfake technology was invented in the late 1990s, but became a widespread phenomenon in 2017. As a result, researchers have voiced concerns about the

Deepfakes are a type of falsified media — audio or visual — produced with deep learning. They employ various techniques — such as face swapping or voice conversion — to mimic a target person’s appearance or voice. The technology can be traced back to 1997 when the first

Convolutional neural networks (CNNs) were first introduced in the 1980s. One of the first known examples of CNN was the Time Delay Neural Network (TDNN) developed in 1987 by Alex Waibel and his research team. TDNN was focused on speech recognition, including shift-invariant

Digital face manipulation has recently emerged as a significant threat to biometric systems. Although manipulation of images/photographs — photoshopping — has been a popular practice for many years, video manipulation has been relatively unknown. Video

Deepfake (derived from ‘deep learning’ and ‘fake’) is a falsified synthetic media — video, photo, or audio — which presents a certain action that was not performed by a given person in reality. This technology uses techniques of deep machine learning and Artificial Intelligence (AI) to