#FactCheck - "Deepfake Video Falsely Claims Justin Trudeau Endorses Investment Project”

Introduction

Search engines have become indispensable in our daily lives, allowing us to find information instantly by entering keywords or phrases. Using the prompt "search Google or type a URL" reflects just how seamless this journey to knowledge has become. With millions of searches conducted every second, and Google handling over 6.3 million searches per minute as of 2023 (Statista), one critical question arises: do search engines prioritise results based on user preferences and past behaviours, or are they truly unbiased?

Understanding AI Bias in Search Algorithms

AI bias is also known as machine learning bias or algorithm bias. It refers to the occurrence of biased results due to human biases that deviate from the original training data or AI algorithm which leads to distortion of outputs and creation of potentially harmful outcomes. The sources of this bias are algorithmic bias, data bias and interpretation bias which emerge from user history, geographical data, and even broader societal biases in training data. 

Common biases include excluding certain groups of people from opportunities because of AI bias. In healthcare, underrepresenting data of women or minority groups can skew predictive AI algorithms. While AI helps streamline the automation of resume scanning during a search to help identify ideal candidates, the information requested and answers screened out can result in biased outcomes due to a biased dataset or any other bias in the input data.

Case in Point: Google’s "Helpful" Results and Its Impact 

Google optimises results by analysing user interactions to determine satisfaction with specific types of content. This data-driven approach forms ‘filter bubbles’ by repeatedly displaying content that aligns with a user’s preferences, regardless of factual accuracy. While this can create a more personalised experience, it risks confining users to a limited view, excluding diverse perspectives or alternative viewpoints.

The personal and societal impacts of such biases are significant. At an individual level, filter bubbles can influence decision-making, perceptions, and even mental health. On a societal level, these biases can reinforce stereotypes, polarise opinions, and shape collective narratives. There is also a growing concern that these biases may promote misinformation or limit users’ exposure to diverse perspectives, all stemming from the inherent bias in search algorithms.

Policy Challenges and Regulatory Measures

Regulating emerging technologies like AI, especially in search engine algorithms, presents significant challenges due to their intricate, proprietary nature. Traditional regulatory frameworks struggle to keep up with them as existing laws were not designed to address the nuances of algorithm-driven platforms. Regulatory bodies are pushing for transparency and accountability in AI-powered search algorithms to counter biases and ensure fairness globally. For example, the EU’s Artificial Intelligence Act aims to establish a regulatory framework that will categorise AI systems based on risk and enforces strict standards for transparency, accountability, and fairness, especially for high-risk AI applications, which may include search engines. India has proposed the Digital India Act in 2023 which will define and regulate High-risk AI. 

Efforts include ethical guidelines emphasising fairness, accountability, and transparency in information prioritisation. However, a complex regulatory landscape could hinder market entrants, highlighting the need for adaptable, balanced frameworks that protect user interests without stifling innovation.

CyberPeace Insights

In a world where search engines are gateways to knowledge, ensuring unbiased, accurate, and diverse information access is crucial. True objectivity remains elusive as AI-driven algorithms tend to personalise results based on user preferences and past behaviour, often creating a biased view of the web. Filter bubbles, which reinforce individual perspectives, can obscure factual accuracy and limit exposure to diverse viewpoints. Addressing this bias requires efforts from both users and companies. Users should diversify sources and verify information, while companies should enhance transparency and regularly audit algorithms for biases. Together, these actions can promote a more equitable, accurate, and unbiased search experience for all users. 

References 

• https://www.bbc.com/future/article/20241101-how-online-photos-and-videos-alter-the-way-you-think 
• https://www.bbc.com/future/article/20241031-how-google-tells-you-what-you-want-to-hear
• https://www.ibm.com/topics/ai-bias#:~:text=In%20healthcare%2C%20underrepresenting%20data%20of,can%20skew%20predictive%20AI%20algorithms

‍

Artificial intelligence is revolutionizing industries such as healthcare to finance to influence the decisions that touch the lives of millions daily. However, there is a hidden danger associated with this power: unfair results of AI systems, reinforcement of social inequalities, and distrust of technology. One of the main causes of this issue is training data bias, which appears when the examples on which an AI model is trained are not representative or skewed. To deal with it successfully, this needs a combination of statistical methods, algorithmic design that is mindful of fairness, and robust governance over the AI lifecycle. This article discusses the origin of bias, the ways to reduce it, and the unique position of fairness-conscious algorithms.

‍

Why Bias in Training Data Matters

‍

The bias in AI occurs when the models mirror and reproduce the trends of inequality in the training data. When a dataset has a biased representation of a demographic group or includes historical biases, the model will be trained to make decisions in ways that will harm the group. This is a fact that has a practical implication: prejudiced AI may cause discrimination during the recruitment of employees, lending, and evaluation of criminal risks, as well as various other spheres of social life, thus compromising justice and equity. These problems are not only technical in nature but also require moral principles and a system of governance (E&ICTA).

Bias is not uniform. It may be based on the data itself, the algorithm design, or even the lack of diversity among developers. The bias in data occurs when data does not represent the real world. Algorithm bias may arise when design decisions inadvertently put one group at an unfair advantage over another. Both the interpretation of the model and data collection may be affected by human bias. (MDPI)

‍

Statistical Principles for Reducing Training Data Bias 

‍

Statistical principles are at the core of bias mitigation and they redefine the data-model interaction. These approaches are focused on data preparation, training process adjustment, and model output corrections in such a way that the notion of fairness becomes a quantifiable goal.

‍

Balancing Data Through Re-Sampling and Re-Weighting

‍

Among the aforementioned methods, a fair representation of all the relevant groups in the dataset is one way. This can be achieved by oversampling underrepresented groups and undersampling overrepresented groups. Oversampling gives greater weight to minority examples, whereas re-weighting gives greater weight to under-represented data points in training. The methods minimize the tendency of models to fit to salient patterns and improve coverage among vulnerable groups. (GeeksforGeeks)

‍

Feature Engineering and Data Transformation

‍

The other statistical technique is to convert data characteristics in such a way that sensitive characteristics have a lesser impact on the results. In one example, fair representation learning adjusts the data representation to discourage bias during the untraining of the model. The disparate impact remover adjust technique performs the adjustment of features of the model in such a way that the impact of sensitive features is reduced during learning. (GeeksforGeeks)

‍

Measuring Fairness With Metrics

‍

Statistical fairness measures are used to measure the effectiveness of a model in groups. 

‍

Fairness-Aware Algorithms Explained

‍

Fair algorithms do not simply detect bias. They incorporate fairness goals in model construction and run in three phases including pre-processing, in-processing, and post-processing.

‍

Pre-Processing Techniques

‍

Fairness-aware pre-processing deals with bias prior to the model consuming the information. This involves the following ways:

‍

• Rebalancing training data through sampling and re-weighting training data to address sample imbalances.
• Data augmentation to generate examples of underrepresented groups.
• Feature transformation removes or downplays the impact of sensitive attributes prior to the commencement of training. (IJMRSET)

‍

These methods can be used to guarantee that the model is trained on more balanced data and to reduce the chances of bias transfer between historical data.

‍

In-Processing Techniques

‍

The in-processing techniques alter the learning algorithm. These include:

‍

• Fairness constraints that penalize the model for making biased predictions during training.
• Adversarial debiasing, where a second model is used to ensure that sensitive attributes are not predicted by the learned representations.
• Fair representation learning that modifies internal model representations in favor of 

‍

Post-Processing Techniques

‍

Fairness may be enhanced after training by changing the model outputs. These strategies comprise:

‍

• Threshold adjustments to various groups to meet conditions of fairness, like equalized odds.
• Calibration techniques such that the estimated probabilities are fair indicators of the actual probabilities in groups. (GeeksforGeeks)

‍

Challenges 

‍

Mitigating bias is complex. The statistical bias minimization may at times come at the cost of the model accuracy, and there is a conflict between predictive performance and fairness. The definition of fairness itself is potentially a difficult task because various applications of fairness require various criteria, and various criteria can be conflicting. (MDPI)

Gaining varied and representative data is also a challenge that is experienced because of privacy issues, incomplete records, and a lack of resources. The auditing and reporting done on a continuous basis are needed so that mitigation processes are up to date, as models are continually updated. (E&ICTA)

‍

Why Fairness-Aware Development Matters

‍

The outcomes of the unfair treatment of some groups by AI systems are far-reaching. Discriminatory software in recruitment may support inequality in the workplace. Subjective credit rating may deprive deserving people of opportunities. Unbiased medical forecasts might result in the flawed allocation of medical resources. In both cases, prejudice contravenes the credibility and clouds the greater prospect of AI. (E&ICTA)

Algorithms that are fair and statistical mitigation plans provide a way to create not only powerful AI but also fair and trustworthy AI. They admit that the results of AI systems are social tools whose effects extend across society. Responsible development will necessitate sustained fairness quantification, model adjustment, and upholding human control.

‍

Conclusion 

‍

AI bias is not a technical malfunction. It is a mirror of real-world disparities in data and exaggerated by models. Statistical rigor, wise algorithm design, and readiness to address the trade-offs between fairness and performance are required to reduce training data bias. Fairness-conscious algorithms (which can be implemented in pre-processing, in-processing, or post-processing) are useful in delivering more fair results. As AI is taking part in the most crucial decisions, it is necessary to consider fairness at the beginning to have a system that serves the population in a responsible and fair manner.

‍

References

‍

• Understanding Bias in Artificial Intelligence: Challenges, Impacts, and Mitigation Strategies: E&ICTA, IITK
• Bias and Fairness in Artificial Intelligence: Methods and Mitigation Strategies: JRPS Shodh Sagar 
• Fairness and Bias in Artificial Intelligence: A Brief Survey of Sources, Impacts, and Mitigation Strategies: MDPI 
• Ensuring Fairness in Machine Learning Algorithms: GeeksforGeeks
  Bias and Fairness in Machine Learning Models: A Critical Examination of Ethical Implications: IJMRSET 
• Bias in AI Models: Origins, Impact, and Mitigation Strategies: Preprints
• Bias in Artificial Intelligence and Mitigation Strategies: TCS
• Survey on Machine Learning Biases and Mitigation Techniques: MDPI

‍

Introduction

You must have heard of several techniques of cybercrime up to this point. Many of which we could never have anticipated. Some of these reports are coming from different parts of the country. Where video calls are being utilised to cheat. Through video calls, cybercriminals are making individuals victims of fraud. During this incident, fraudsters film pornographic recordings of both the victims using a screen recorder, then blackmail them by emailing these videos and demanding money. However, cybercriminals are improving their strategies to defraud more people. In this blog post, we will explore the tactics involved in this case, the psychological impact, and ways to combat it. Before we know more about the case, let’s have a look at deep fake, AI, and Sextortion and how fraudsters use technology to commit crimes.

Understanding Deepfake

Deepfake technology is the manipulation or fabrication of multimedia information such as videos, photos, or audio recordings using artificial intelligence (AI) algorithms, and profound learning models. These algorithms process massive quantities of data to learn and imitate human-like behaviour, allowing for very realistic synthetic media development.

Individuals with malicious intent may change facial expressions, bodily movements, and even voices in recordings using deepfake technology, basically replacing a person’s appearance with someone else’s. The produced film can be practically indistinguishable from authentic footage, making it difficult for viewers to distinguish between the two.

Sextortion and technology

Sextortion is a sort of internet blackmail in which offenders use graphic or compromising content to compel others into offering money, sexual favours, or other concessions. This information is usually gained by hacking, social engineering, or tricking people into providing sensitive information.

Deepfake technology combined with sextortion techniques has increased the impact on victims. Deepfakes may now be used by perpetrators to make and distribute pornographic or compromising movies or photographs that seem genuine but are completely fake. As the prospect of discovery grows increasingly credible and tougher to rebut, the stakes for victims rise.

Cyber crooks Deceive

In this present case, cyber thugs first make video calls to people and capture the footage. They then twist the footage and merge it with a distorted naked video. As a result, the victim is obliged to conceal the case. Following that, “they demand money as a ransom to stop releasing the doctored video on the victim’s contacts and social media platforms.” In this case, a video has emerged in which a lady who was supposedly featured in the first film is depicted committing herself because of the shame caused by the video’s release. These extra threats are merely intended to inflict psychological pressure and coercion on the victims.

Sextortionists have reached a new low by profiting from the misfortunes of others, notably targeting deceased victims. The offenders want to maximise emotional pain and persuade the victim into acquiescence by generating deep fake films depicting these persons. They use the inherent compassion and emotion connected with tragedy to exact bigger ransoms from their victims.

This distressing exploitation not only adds urgency to the extortion demands but also preys on the victim’s sensitivity and emotional instability. They even pressurize the victim by impersonating them, and if the demands are fulfilled, the victims may land up in jail.

Tactics used

The morphed death videos are precisely constructed to heighten emotional discomfort and instil terror in the targeted individual. By editing photographs or videos of the deceased, the offenders create unsettling circumstances that heighten the victim’s emotional response.

The psychological manipulation seeks to instil guilt, regret, and a sense of responsibility in the victim. The notion that they are somehow linked to the catastrophe increases their emotional weakness, making them more vulnerable to the demands of sextortionists. The offenders take use of these emotions, coercing victims into cooperation out of fear of being involved in the apparent tragedy.

The impact on the victim’s mental well-being cannot be overstated. They may experience intense psychological trauma, including anxiety, depression, and post-traumatic stress disorder (PTSD). The guilt and shame associated with the false belief of being linked to someone’s death can have long-lasting effects on their emotional health and overall quality of life, others may have trust issues.

Law enforcement agencies advised

Law enforcement organisations were concerned about the growing annoyance of these illegal acts. The use of deep fake methods or other AI technologies to make convincing morphing films demonstrates scammers’ improved ability. These tools are fully capable of modifying digital information in ways that are radically different from the genuine film, making it difficult for victims to detect the fake nature of the video.

Defence strategies to fight back: To combat sextortion, a proactive approach that empowers individuals and utilizes resources is required. This section delves into crucial anti-sextortion techniques such as reporting events, preserving evidence, raising awareness, and implementing digital security measures.

• Report the Incident: Sextortion victims should immediately notify law enforcement. Contact your local police or cybercrime department and supply them with any important information, including specifics of the extortion attempt, communication logs, and any other evidence that can assist in the investigation. Reporting the occurrence is critical for keeping criminals responsible and averting additional harm to others.
• Preserve Evidence: Preserving evidence is critical in creating a solid case against sextortionists. Save and document any types of contact connected to the extortion, including text messages, emails, and social media conversations. Take screenshots, record phone calls (if legal), and save any other digital material or papers that might be used as evidence. This evidence can be useful in investigations and judicial processes.

Digital security: Implementing comprehensive digital security measures can considerably lower the vulnerability to sextortion assaults. Some important measures that one can use:

• Use unique, complicated passwords for all online accounts, and avoid reusing passwords across platforms. Consider utilising password managers to securely store and create strong passwords.
• Enable two-factor authentication (2FA) whenever possible, which adds an extra layer of protection by requiring a second verification step, such as a code delivered to your phone or email, in addition to the password.
• Regular software updates: Keep your operating system, antivirus software, and programmes up to date. Security patches are frequently included in software upgrades to defend against known vulnerabilities.
• Adjust your privacy settings on social networking platforms and other online accounts to limit the availability of personal information and restrict access to your content.
• Be cautious when clicking on links or downloading files from unfamiliar or suspect sources. When exchanging personal information online, only use trusted websites.

Conclusion:

Combating sextortion demands a collaborative effort that combines proactive tactics and resources to confront this damaging practice. Individuals may actively fight back against sextortion by reporting incidences, preserving evidence, raising awareness, and implementing digital security measures. It is critical to empower victims, encourage their rehabilitation, and collaborate to build a safer online environment where sextortionists are held accountable and everyone can navigate the digital environment with confidence.