#FactCheck-Old Bihar train protest video falsely linked to CJP protest and India’s first hydrogen train.

Introduction

In 2019 India got its bill on Data protection in the form of the Personal Data Protection Bill 2019. This bill focused on digital rights and duties pertaining to data privacy. However, the bill was scrapped by the Govt in mid-2022, and a new bill was drafted, Successor bill was introduced as the Digital Personal Data Protection Bill, 2022 on 18th November 2022, which was made open for public comments and consultations and now the bill is expected to be tabled at the parliament in the Monsoon session.

What is DPDP, 2022?

Digital Personal Data Protection Bill, is the lasted draft regulation for data privacy in India. The bill has been essentially focused towards data protection by companies and the keep aspect of Puttaswamy judgement of data privacy as a fundamental right has been upheld under the scope of the bill. The bill comes after nearly 150 recommendations which the parliamentary committee made when the PDP, 2019 was scrapped.

The bill highlights the following keen aspects-

• Data Fiduciary- The entity (an individual, company, firm, state, etc.) which decides the purpose and means of processing an individual’s personal data.
• Data Principle- The individual to whom personal data is related.
• Processing- The entire cycle of operations that can be carried out concerning personal data.
• Gender Neutrality- For the first time in India’s legislative history, “her” and “she” have been used to refer to individuals irrespective of gender.
• Right to Erase Data- Data principals will have the right to demand the erasure and correction of data collected by the data fiduciary.
• Cross-border data transfer- The bill allows cross-border data after an assessment of relevant factors by the Central Government.
• Children’s Rights- The bill guarantees the right to digital privacy under the protection of parents/guardians.
• Heavy Penalties- The bill enforces heavy penalties for non-compliance with the provisions, not exceeding Rs 500 crore.

Data Protection Board

The bill lays down provisions for setting up a Data Protection Board. This board will be an independent body acting solely on the factors of data privacy and protection of the data principles and maintaining compliance by data fiduciaries. The board will be headed by a chairperson of essential and relevant qualifications, and members and various other officials shall assist him/her under the board. The board will serve grievance redressal to the data principles and can conduct investigation, inquiry, proceeding, and pass orders equivalent to a Civil court. The proceeding will be undertaken on the principle of natural justice, and the aggrieved can file an appeal to the High Court of appropriate jurisdiction.

Global Comparison

Many countries have data protection laws that regulate the processing of personal data. Some of the notable examples include:

• European Union: The EU’s General Data Protection Regulation (GDPR) is one of the world’s most comprehensive data protection laws. It regulates public and private entities’ processing of personal data and gives individuals a wide range of rights over their personal data.
• United States: The US has several data protection laws that apply to specific sectors or types of data, such as health data (HIPAA) or financial data (Gramm-Leach-Bliley Act). However, there is no comprehensive federal data protection law in the US.
• Japan: Japan’s Personal Information Protection Act (PIPA) regulates the handling of personal data by private entities and gives individuals certain rights over their personal data.
• Australia: Australia’s Privacy Act 1988 regulates the handling of personal data by public and private entities and gives individuals certain rights over their personal data.
• Brazil: Brazil’s General Data Protection Law (LGPD) regulates the processing of personal data by public and private entities and gives individuals certain rights over their personal data. It also imposes heavy fines and penalties on entities that violate the provisions of the law.

Overall, while there are some similarities in data protection laws across countries, there are also significant differences in scope, applicability, and enforcement. It is important for organisations to understand the data protection laws that apply to their operations and take appropriate steps to comply with these laws.

Parliamentary Asscent

The case of violation of the privacy policy by WhatsApp at the Hon’ble Supreme Court resulted in a significant advocacy for Data privacy as a fundamental right, and it was held that, as suggested otherwise in the privacy policy, Whatsapp was sharing its user’s data with Meta. This massive breach of trust could have led to data mismanagement affecting thousands of Indian users. The Hon’ble Supreme Court has taken due consideration of data privacy and its challenges in India and asked the Govt to table the bill in Parliament. The bill will be tabled for discussion in the monsoon session. The Supreme Court has set up a constitutional bench to check the bill’s scope, extent and applications and provide its judicial oversight.  The constitution bench of Justices KM Joseph, Ajay Rastogi, Aniruddha Bose, Hrishikesh Roy and CT Ravikumar has fixed the matter for hearing in August in order to enforce the potential changes and amendments in the act post the parliamentary discussion.

Conclusion

India is the world’s largest democracy, so the crucial aspects of passing laws and amendments have always been followed by the government and kept under check by the judiciary. The discussion over bills is a crucial part of the democratic process, and bills as important as Digital Personal Data Protection need to be discussed and analysed thoroughly in both houses of Parliament to ensure the govt passes a sustainable and efficient law.

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

‍

The use of AI in content production, especially images and videos, is changing the foundations of evidence. AI-generated videos and images can mirror a person’s facial features, voice, or actions with a level of fidelity to which the average individual may not be able to distinguish real from fake. The ability to provide creative solutions is indeed a beneficial aspect of this technology. However, its misuse has been rapidly escalating over recent years. This creates threats to privacy and dignity, and facilitates the creation of dis/misinformation. Its real-world consequences are the manipulation of elections, national security threats, and the erosion of trust in society. 

‍

Why India Needs Deepfake Regulation

‍

Deepfake regulation is urgently needed in India, evidenced by the recent Rashmika Mandanna incident, where a hoax deepfake of an actress created a scandal throughout the country. This was the first time that an individual's image was superimposed on the body of another woman in a viral deepfake video that fooled many viewers and created outrage among those who were deceived by the video. The incident even led to law enforcement agencies issuing warnings to the public about the dangers of manipulated media.

This was not an isolated incident; many influencers, actors, leaders and common people have fallen victim to deepfake pornography, deepfake speech scams, defraudations, and other malicious uses of deepfake technology. The rapid proliferation of deepfake technology is outpacing any efforts by lawmakers to regulate its widespread use. In this regard, a Private Member’s Bill was introduced in the Lok Sabha in its Winter Session. This proposal was presented to the Lok Sabha as an individual MP's Private Member's Bill. Even though these have had a low rate of success in being passed into law historically, they do provide an opportunity for the government to take notice of and respond to emerging issues. In fact, Private Member's Bills have been the catalyst for government action on many important matters and have also provided an avenue for parliamentary discussion and future policy creation. The introduction of this Bill demonstrates the importance of addressing the public concern surrounding digital impersonation and demonstrates that the Parliament acknowledges digital deepfakes to be a significant concern and, therefore, in need of a legislative framework to combat them.

‍

Key Features Proposed by the New Deepfake Regulation Bill

‍

The proposed legislation aims to create a strong legal structure around the creation, distribution and use of deepfake content in India. Its five core proposals  are: 

1. Prior Consent Requirement: individuals must give their written approval before producing or distributing deepfake media, including digital representations of themselves, as well as their faces, images, likenesses and voices. This aims to protect women, celebrities, minors, and everyday citizens against the use of their identities with the intent to harm them or their reputations or to harass them through the production of deepfakes.

2. Penalties for Malicious Deepfakes: Serious criminal consequences should be placed for creating or sharing deepfake media, particularly when it is intended to cause harm (defame, harass, impersonate, deceive or manipulate another person). The Bill also addresses financially fraudulent use of deepfakes, political misinformation, interfering with elections and other types of explicit AI-generated media. 

3. Establishment of a Deepfake Task Force: To look at the potential impact of deepfakes on national security, elections and public order, as well as on public safety and privacy. This group will work with academic institutions, AI research labs and technology companies to create advanced tools for the detection of deepfakes and establish best practices for the safe and responsible use of generative AI. 

4. Creation of a Deepfake Detection and Awareness Fund: To assist with the development of tools for detecting deepfakes, increasing the capacity of law enforcement agencies to investigate cybercrime, promoting public awareness of deepfakes through national campaigns, and funding research on artificial intelligence safety and misinformation.

‍

How Other Countries Are Handling Deepfakes

‍

1. United States
Many States in the United States, including California and Texas, have enacted laws to prohibit the use of politically deceptive deepfakes during elections. Additionally, the Federal Government is currently developing regulations requiring that AI-generated content be clearly labelled. Social Media Platforms are also being encouraged to implement a requirement for users to disclose deepfakes.

2. United Kingdom
In the United Kingdom, it is illegal to create or distribute intimate deepfake images without consent; violators face jail time. The Online Safety Act emphasises the accountability of digital media providers by requiring them to identify, eliminate, and avert harmful synthetic content, which makes their role in curating safe environments all the more important.

3. European Union:
The EU has enacted the EU AI Act, which governs the use of deepfakes by requiring an explicit label to be affixed to any AI-generated content. The absence of a label would subject an offending party to potentially severe regulatory consequences; therefore, any platform wishing to do business in the EU should evaluate the risks associated with deepfakes and adhere strictly to the EU's guidelines for transparency regarding manipulated media.

4. China:
China has among the most rigorous regulations regarding deepfakes anywhere on the planet. All AI-manipulated media will have to be marked with a visible watermark, users will have to authenticate their identities prior to being allowed to use advanced AI tools, and online platforms have a legal requirement to take proactive measures to identify and remove synthetic materials from circulation.

‍

Conclusion

‍

Deepfake technology has the potential to be one of the greatest (and most dangerous) innovations of AI technology. There is much to learn from incidents such as that involving Rashmika Mandanna, as well as the proliferation of deepfake technology that abuses globally, demonstrating how easily truth can be altered in the digital realm. The new Private Member's Bill created by India seeks to provide for a comprehensive framework to address these abuses based on prior consent, penalties that actually work, technical preparedness, and public education/awareness. With other nations of the world moving towards increased regulation of AI technology, proposals such as this provide a direction for India to become a leader in the field of responsible digital governance.

‍

References 

‍

1. https://www.ndtv.com/india-news/lok-sabha-introduces-bill-to-regulate-deepfake-content-with-consent-rules-9761943
2. https://m.economictimes.com/news/india/shiv-sena-mp-introduces-private-members-bill-to-regulate-deepfakes/articleshow/125802794.cms
3. https://www.bbc.com/news/world-asia-india-67305557
4. https://www.akingump.com/en/insights/blogs/ag-data-dive/california-deepfake-laws-first-in-country-to-take-effect
5. https://codes.findlaw.com/tx/penal-code/penal-sect-21-165/ 
6. https://www.mishcon.com/news/when-ai-impersonates-taking-action-against-deepfakes-in-the-uk#:~:text=As%20of%2031%20January%202024,of%20intimate%20deepfakes%20without%20consent. 
7. https://www.politico.eu/article/eu-tech-ai-deepfakes-labeling-rules-images-elections-iti-c2pa/ 
8. https://www.reuters.com/article/technology/china-seeks-to-root-out-fake-news-and-deepfakes-with-new-online-content-rules-idUSKBN1Y30VT/

‍