Decoding Deepfakes: Understanding the Threat and its Implications

A viral deepfake video of actress Rashmika Mandanna has highlighted the unsettling world of AI deepfakes and the perils of unregulated access of the growing Artificial Intelligence (AI). However, Rashmika is not the only celebrity who has fallen victim to deepfakes; their misuse is escalating daily, with trolls weaponizing them to harm individuals’ sentiments.

The video depicts a woman with Rashmika’s face, clad in a black fitted outfit, entering a lift. The face has been manipulated to resemble Mandanna and has garnered over 12 million views on X (formerly Twitter). This incident underscores the increasing misuse of deepfakes, with trolls exploiting them as a potent weapon to inflict emotional harm.

In this blog, we will unravel the concept of deepfakes, explore why they pose a significant threat, and provide insights on safeguarding yourself from potential harm. Understanding the depths of this deceptive technology is crucial in navigating the evolving landscape of AI and protecting against its misuse.

Also Read: How To Earn $1,000/Month With The SHEIN Affiliate Program In 2023

What is Deepfakes?

Deepfakes, powered by Artificial Intelligence (AI), craft convincing fake content like images and videos. They use tools such as AI, photoshop, and machine learning, creating fake videos that seem real. These can feature fictional photos, morphed videos, or even ‘voice clones’ of public figures.

To make a deepfake video look authentic, AI algorithms need substantial training data, often sourced from social media, containing various recordings of the target’s voice. The more diverse the dataset, the more convincing the replicated voice becomes.

Deepfakes go beyond traditional photo manipulation. They often swap faces in existing content or generate entirely new content, showing individuals doing or saying things they never did. The synthetic medium, fueled by AI and Machine Learning, poses risks in a world already dealing with misinformation and manipulated truths.

Though not a new concept, deepfakes now pose a serious threat to privacy and truth. Their easy creation using available technology amplifies risks, especially in contexts like India, where misinformation can be exploited for political manipulation.

Also Read: Amitabh Bachchan and Flipkart: All You Need to Know About the Legal Dispute Over Misleading Ads

What was Rashmika Deepfakes Scandal?

The Rashmika Mandanna deepfake scandal unfolded as a viral video showcased an eerily realistic transformation, seamlessly morphing an unknown woman’s face into that of Rashmika Mandanna. The video, capturing Mandanna entering an elevator in a black dress, sparked a conversation about the dark side of Artificial Intelligence (AI) and the growing threat of deepfake videos.

The woman featured in the video is identified as British Indian social media personality Zara Patel, who clarified her non-involvement in the controversy through a statement on social media, expressing deep upset about the situation.

Also Read: ART-E launches ‘Aayega Teesra’ Anthem, Paving the Path to Team India’s 3rd World Cup

Rashmika Mandanna, upon discovering the deepfake video, expressed her deep concern, labeling it as “extremely scary.” She emphasized the broader vulnerability of individuals to harm due to the misuse of technology.

I feel really hurt to share this and have to talk about the deepfake video of me being spread online.

Something like this is honestly, extremely scary not only for me, but also for each one of us who today is vulnerable to so much harm because of how technology is being misused.…

— Rashmika Mandanna (@iamRashmika) November 6, 2023

This incident transcended the entertainment realm, prompting widespread concern and calls for immediate legal action. Bollywood icon Amitabh Bachchan, Mandanna’s co-star from the movie “Goodbye,” raised his voice against the dangers of deepfakes and stressed the necessity for legal intervention.

yes this is a strong case for legal https://t.co/wHJl7PSYPN

— Amitabh Bachchan (@SrBachchan) November 5, 2023

Union Minister Rajeev Chandrasekhar supported Rashmika, reminding social media platforms of their legal obligations in combatting misinformation. He underscored the government’s commitment to ensuring the safety and trust of all digital users on the internet.

PM @narendramodi ji's Govt is committed to ensuring Safety and Trust of all DigitalNagriks using Internet

Under the IT rules notified in April, 2023 – it is a legal obligation for platforms to

➡️ensure no misinformation is posted by any user AND

➡️ensure that when reported by… https://t.co/IlLlKEOjtd

— Rajeev Chandrasekhar 🇮🇳 (@Rajeev_GoI) November 6, 2023

The Rashmika Mandanna deepfake controversy is not an isolated incident; numerous celebrities have fallen victim to deepfake manipulations. Recently, Katrina Kaif also became a victim of deepfake, where a scene from Tiger 3 was manipulated by deepfake technology, adding to the growing list of celebrities affected by this concerning trend. Actor Miles Fisher, resembling Tom Cruise, gained internet fame with deepfake TikTok videos, while viral Emma Watson lookalike videos turned out to be deepfake creations. A-list celebrities like Jennifer Lawrence, Arnold Schwarzenegger, and tech mogul Mark Zuckerberg have also faced dubious deepfake videos. The prevalence of such manipulations raises concerns about the potential misuse of deepfake technology across various platforms.

Also Read: Decoding Flipkart Innovative Marketing during Big Billion Days

Creating deepfakes involves a complex process, primarily driven by machine learning. A common misconception is that generative adversarial networks (GANs) are the main engine behind deepfake development. While GANs are powerful, other algorithms, both AI and non-AI, play crucial roles in the creation of deepfake videos.

Types of Deepfakes

Types of Deepfakes: Exploring the Diverse Landscape

1. Textual Deepfakes

• In the realm of evolving AI, textual deepfakes have emerged, generating human-like written content such as articles and poems.
• Examples include GPT-2 and GPT-3 by OpenAI, capable of crafting stories and news articles indistinguishable from human writing.

Also Read: Are AI-Generated Influencers the Future of Marketing?

2. Deepfake Videos

• Among the most prevalent types, deepfake videos employ AI and advanced video editing to create realistic yet fabricated content.
• With ample images, various software enables the substitution of one person’s face in a video with another, often mimicking celebrities or public figures.

3. Deepfake Images

• Easily crafted with basic skills and accessible online, deepfake images involve superimposing faces onto different bodies.
• Editing software facilitates the creation of highly realistic images, raising concerns about their potential misuse.

Also Read: The 15 Top Artists on Spotify Worldwide In 2023

4. Deepfake Audio

• Advancements in deepfake technology extend to synthetic audio content, replicating a person’s voice, tone, and accent.
• Algorithms analyze existing voice clips to generate new ones, with the credibility increasing with the availability of more voice recordings.

5. Real-Time Deepfakes

• The pinnacle of deepfake evolution, real-time or live deepfakes recreate various content forms instantaneously.
• Platforms like DeepFaceLive exemplify this, enabling users to impersonate someone during live streams or video calls, posing unique challenges for authentication and trust.

Deepfake Creation Process

1. Data Collection

• Gather many hours of real video footage featuring the target person.
• Train a neural network on this data to develop a realistic understanding of the person’s appearance from various angles and under different lighting conditions.

2. Neural Network Training

• Use machine learning to train a neural network to comprehend the facial features, expressions, and movements of the target person.
• This step enables the algorithm to create a digital representation of the person’s likeness.

3. Combination with Computer Graphics

• Combine the trained neural network with computer graphics techniques to overlay the digital representation onto a different actor.
• This process involves superimposing the target person’s features onto a new context, creating a deepfake.

Also Read: Ayushmann Khurrana Partners With Agoda For Its First-Ever Indian TV Ad

4. Manual Adjustment

• Manually adjust the parameters of the trained program to eliminate telltale signs, blips, or artifacts that may reveal the artificial nature of the deepfake.
• This step is crucial for enhancing the overall believability of the composite.

5. Role of GANs

• While GANs are often associated with deepfakes, they are not the sole driving force.
• GANs are powerful for synthesizing images but face challenges with temporal consistency in videos, making them less suitable for creating deepfake videos.

Also Read: The 12 Best Superhero Movies To Watch on Netflix For Every Fan

6. Diversity in Algorithms

• Deepfakes rely on a constellation of AI and non-AI algorithms.
• GANs are challenging to work with and demand substantial training data, slowing down the image generation process.
• Many deepfake videos today are created using a combination of different algorithms to achieve the desired results.

In essence, the creation of deepfakes involves a multidimensional process, where machine learning, neural networks, and various algorithms work together to generate convincing but misleading content. The complexity of the task necessitates a careful balance of different techniques to achieve a seamless and realistic deepfake.

How deep fake videos are being misused

Misuse of Deepfake Videos

1. Non-consensual Pornography:
   • Predominantly used for creating explicit content without consent, targeting women specifically.
   • Rising cases of non-consensual deepfake porn involving celebrities like Taylor Swift and Emma Watson.
2. Privacy Invasion and Identity Theft:
   • Deepfakes contribute to the invasion of personal privacy, as individuals’ faces are superimposed onto explicit or misleading content.
   • Heightened risk of identity theft as facial features are manipulated for malicious purposes.
3. Cyberbullying and Harassment:
   • Deepfakes are employed as tools for cyberbullying, harassment, and intimidation.
   • Perpetrators use manipulated videos or images to target individuals online, causing emotional distress.
4. Fraud and Scams:
   • Fraudsters employ deepfakes to create fake personas for romance scams, emotionally manipulating victims online.
   • Cases where individuals have been deceived, resulting in significant financial losses.
5. Commercial Exploitation:
   • Deepfakes can be used for unauthorized commercial exploitation, incorporating individuals’ images into advertising campaigns or movies without consent.
   • Example: “Joan Is Awful” episode in the Black Mirror anthology series.
6. Impersonation:
   • Perpetrators use deepfakes to impersonate people, including family members and colleagues of their victims.
   • Creates confusion and can lead to misunderstandings or harm relationships.
7. Fake Recordings and Spreading False Information:
   • Deepfakes are utilized to create fake customer service recordings or impersonate public figures.
   • Spread false information, contributing to the dissemination of misinformation and deception.
8. Manipulative Use in Romance Scams:
   • Deepfakes are employed to create deceptive personas for online romance scams, emotionally manipulating victims.
9. Copyright Issues:
   • Unauthorized use of someone’s likeness raises copyright concerns, as individuals have little control over how their images are manipulated and disseminated.
10. Gendered Violence:
    • Deepfakes contribute to technology-facilitated online gendered violence, particularly against women.
    • A significant percentage of deepfake content is sexually explicit, often used to perpetuate harm.

Also Read: Saumya Tandon Becomes the Brand Ambassador of MamyPoko Pants In Uttar Pradesh

In summary, deepfake videos are increasingly being exploited for malicious purposes, ranging from privacy invasion and identity theft to cyberbullying, fraud, and the spread of false information. The rise of deepfake technology poses significant challenges in addressing these issues and protecting individuals from various forms of exploitation and harm.

How to identify a deep fakes Video

1. Strange Looking Features:
   • Visual signs like unusual artefacts, inconsistent facial features, or unnatural movements.
   • Look for video frames that appear blurry or hazy.
2. Biological and Physiological Signs:
   • Abnormal blinking frequency, eye color variations, or irregular heartbeat patterns can be indicative of a deepfake.
   • Physiological signals can help assess the authenticity of the video.
3. Audio-Visual Fallacies:
   • Analyze both the audio and visual aspects for inconsistencies.
   • Mismatch between audio and mouth movements can signal a deepfake.
4. Identity Information:
   • Unique patterns in speech, facial expressions, and gestures for each individual.
   • Detection of inconsistencies or strange patterns in these aspects may reveal a deepfake.
5. Facial Emotions:
   • Unnatural displays of emotions, facial expressions, or vocalizations.
   • Analysis of emotional cues can help determine the authenticity of the video.
6. Temporal Inconsistencies:
   • Look for temporal inconsistencies, where the video lacks smooth flow or exhibits sudden changes in facial positions.
   • Detection of these irregularities can be a strong indicator of a deepfake.
7. Spatial-Temporal Features:
   • Examination of both spatial (visual) and temporal (time-related) features.
   • Abnormalities or inconsistencies in these features may reveal manipulation.

Remember, the identification of deepfake videos often involves a combination of these factors. While technology for creating deepfakes continues to advance, so does the technology for detecting them. Staying vigilant and employing a critical eye can contribute to the identification and prevention of the spread of deceptive deepfake content.

Laws Against Deepfake Fraud

1. Information Technology Act (IT Act):
   • Section 66D of the IT Act, 2000 targets the misuse of communication devices for cheating or impersonation, carrying penalties of imprisonment for up to three years and a fine of Rs 1 lakh.
2. Protection of Privacy – Section 66E of IT Act:
   • Section 66E addresses the breach of privacy through the capture, publication, or transmission of images on the internet.
   • Offenders can face imprisonment for up to three years or a fine of up to Rs 2 lakh.
3. Copyright Act, 1957 – Section 51:
   • Section 51 of the Copyright Act identifies the violation when another person’s exclusive rights are infringed.
   • Deepfake content using someone’s likeness without consent may fall under this provision.
4. Data Protection Bill 2021:
   • The Data Protection Bill 2021 contains provisions to penalize the breach of personal and non-personal data, addressing cybercrimes like deep fakes.
5. IT Rules 2023:
   • IT Amendment Rules, 2023 make it a legal obligation for digital platforms to prevent the posting of misinformation.
   • Platforms must ensure the removal of misinformation within 36 hours after reporting by users or government authorities.

Legal Measures Against Deepfake Media in India

Privacy Laws: The Information Technology Act and its regulations safeguard an individual’s privacy, granting the right to data privacy. If a deepfake video infringes on an individual’s privacy by using their likeness without consent, the victim can file a complaint under this law. Section 66 D of the IT Act addresses punishment for cheating by personation through a computer resource, with penalties of imprisonment up to three years and a fine of up to Rs 1 lakh.

Responsibility on Social Media Platforms: Under the Information Technology Intermediary Rules, social media platforms are mandated to observe due diligence. Rule 3(1)(b)(vii) emphasizes the responsibility of platforms to inform users not to host content impersonating another person. Rule 3(2)(b) further requires prompt action, within 24 hours of receiving a complaint, to remove or disable access to content involving impersonation or artificially morphed images.

Defamation Laws: The Indian Penal Code (IPC) contains provisions for defamation (Sections 499 and 500). Victims of deepfake videos intended to harm reputations through false information can file defamation suits. Establishing defamation in the context of deepfakes requires proving falsity, publication to a third party, harm to the subject’s reputation, and fault on the part of the video’s creator.

Cybercrime: The Information Technology Act and its rules cover a broad range of cybercrimes, including unauthorized access, data theft, and cyberbullying. Victims of deepfake videos produced through illicit methods like hacking or data theft can file complaints under this law, addressing offenses involving unauthorized access to computer resources and potential breaches of sensitive personal data protection.

Right to be Forgotten: While India lacks a specific “right to be forgotten” law, individuals can approach courts to request the removal of personal information, including deepfake videos, from the internet. Courts may consider such requests based on privacy and data protection principles.

Consumer Protection Laws: The Consumer Protection Act of 2019 provides a legal recourse for individuals affected by the creation or distribution of deepfake content used for fraudulent purposes. This law aims to protect consumers’ rights and interests in cases of fraud or misrepresentation.

While India may not have a specific law explicitly addressing deepfake cybercrimes, existing legislation provides avenues for legal action against offenders. The evolving legal landscape, including pending bills and amendments, indicates a growing recognition of the need to combat deepfake-related challenges and protect individuals’ rights.

How can you prevent yourself from Deepfake Fraud?

1. Limit Personal Information:
   • Tip: Reduce the amount of personal information shared online.
   • Explanation: Minimize the data available for creating deepfakes by being mindful of the information shared on social media and other online platforms.
2. Enhance Personal Security:
   • Tip: Strengthen your personal security.
   • Explanation: Refrain from sharing photos or videos that could potentially be used for creating deepfakes. Be cautious about the content you post online.
3. Adjust Instagram Privacy Settings:
   • Tip: Consider changing your Instagram account to private.
   • Explanation: If you are concerned about personal data, setting your Instagram account to private limits access to your content to approved followers, reducing the risk of misuse.
4. Hide Personal Images on Business Accounts:
   • Tip: If using a business account, hide personal images and videos on Instagram.
   • Explanation: Business accounts often share a mix of personal and professional content. Hiding personal images adds an extra layer of privacy.
5. Exercise Caution on Social Media:
   • Tip: Exercise caution while using social media.
   • Explanation: Be mindful of the information you share and the authenticity of content you encounter. Avoid engaging with suspicious or unfamiliar profiles.
6. Secure Social Media Handles:
   • Tip: Keep your social media handles secure.
   • Explanation: Use strong, unique passwords and enable two-factor authentication to prevent unauthorized access to your accounts.

These basic tips provide a foundation for protecting yourself against deepfake fraud. By being vigilant about personal information, adjusting privacy settings, and exercising caution online, individuals can reduce the risk of falling victim to deepfake-related incidents.

Conclusion

In the wake of the Rashmika Mandanna deepfake scandal, the blog illuminates the pervasive threat of deepfakes, showcasing their potential for harm across diverse sectors. From the intricate process of their creation to the legal fortifications against their malicious use in India, the narrative underscores the need for heightened awareness, technological vigilance, and legal safeguards to protect against the growing menace of digital deception.

As individuals navigate the digital realm, the blog advocates for proactive measures, emphasizing the importance of limiting personal information, bolstering social media security, and staying vigilant against the multifaceted dangers posed by deepfakes. In this collective effort, readers are urged to champion digital integrity and contribute to the ongoing battle against the deceptive shadows cast by evolving technology.

FAQ

Q1: What is a deepfake?

Deepfake refers to manipulated media, often videos, created using artificial intelligence (AI) and machine learning techniques to depict individuals saying or doing things they never did.

Q2: How are deepfakes created?

Deepfakes are generated using AI algorithms that analyze and mimic human voices and appearances. Training data, often from social media, helps create realistic replicas.

Q3: Why are deepfakes considered dangerous?

Deepfakes pose threats to privacy, misinformation, and reputation harm. They can be used to create false narratives, manipulate public opinion, and even generate non-consensual explicit content.

Q4: Have celebrities been victims of deepfake videos?

Yes, celebrities like Rashmika Mandanna, Miles Fisher, Emma Watson, Jennifer Lawrence, Arnold Schwarzenegger, Mark Zuckerberg, and recently, Katrina Kaif, have fallen victim to deepfake manipulations.

Q5: How can one protect themselves from deepfake fraud?

Limit personal information online, be cautious on social media, and consider changing privacy settings. If a victim, explore legal avenues under privacy laws, defamation, and cybercrime statutes.

Q6: What laws exist in India to combat deepfake fraud?

Privacy laws, defamation laws (IPC Sections 499 and 500), the Information Technology Act (Sections 66 D and 66E), the Copyright Act, Data Protection Bill 2021, and Consumer Protection Act of 2019 can be invoked against deepfake offenses.

Q7: Can social media platforms be held accountable for hosting deepfake content? Yes, under Information Technology Intermediary Rules, platforms must ensure due diligence and promptly remove or disable access to content involving impersonation or deepfakes.

Q8: Is there a “right to be forgotten” in India for deepfake removal?

While not explicitly, individuals can approach courts based on privacy and data protection principles to request the removal of personal information, including deepfakes, from the internet.

Q9: What role do consumer protection laws play in addressing deepfake-related fraud?

The Consumer Protection Act of 2019 provides relief for individuals affected by deepfake content used for fraudulent purposes, aiming to protect consumers’ rights and interests.