Received: 13 November 2025; Revised: 25 December 2025; Accepted: 25 December 2025; Published Online: 27 December 2025.

J. Collect. Sci. Sustain., 2025, 1(3), 25411 | Volume 1 Issue 3 (December 2025) | DOI: https://doi.org/10.64189/css.25411

This article is licensed under Creative Commons Attribution NonCommercial 4.0 International (CC-BY-NC 4.0)

Responsible AI in the Age of Generative Media: A

Comparative Study of Ethical and Transparent AI

Frameworks in India and the USA

Bhavya S. Tripathi,

1,*

Shrawani C. Gaonkar

and Anagha Dhavalikar

Department of Artificial Intelligence and Data Science, Vasantdada Patil Pratishthan’s College of Engineering & Visual Arts, Mumbai,

Maharashtra, 400022, India

Department of Electronics and Computer Science, Vasantdada Patil Pratishthan’s College of Engineering & Visual Arts, Mumbai,

Maharashtra, 400022, India

*Email: bhavyatripathi02@gmail.com (B. S. Tripathi)

Abstract

Advances in generative artificial intelligence, such as Google Gemini, ChatGPT and DALL·E, are opening new

possibilities for creativity on digital media but also raising pressing concerns about misinformation, deepfakes and

declining trust in media outlets. This paper explores Responsible Artificial Intelligence (RAI) efforts at the policy level

in India and the United States, comparing their unique and shared approaches to addressing generative media -

where policies are driven toward balancing innovation with transparency, accountability and fairness. India's official

AI for All strategy speaks to inclusivity and social development, though concrete enforcement mechanisms and

safeguards against generative media misuse are currently lacking. The US relies on the National Institute of Standards

and Technology (NIST) AI Risk Management Framework, emphasizing risk assessment, technical robustness and

accountability but lacks a robust regulatory mechanism that aligns its varying state and sector-specific initiatives.

Alongside policy analysis, we designed and ran a pilot survey of college students and working professionals through

September 2025 to capture awareness, trust and concerns related to Responsible AI. At a preliminary level, our

findings showed low relative exposure to national policy frameworks on RAI but high expressed concern related to

potential misuse of generative AI around deepfakes, manipulated images and the lack of checks on content

authenticity. Respondents expressed particularly high endorsement for content labels that would mandate labeling

artificial intelligence generated media. We conclude that India and the USA display parallel and diverging paths on

RAI for generative media but both experiences are marked by a gap between policy aspirations and public

understanding. Moving ahead, we see a need for greater policy clarity, cross-border coordination and public outreach

to foster transparency and responsibility in adoption of AI for media.

Keywords: Generative artificial intelligence; Responsible AI; Ethical AI adoption; AI transparency and accountability;

Policy frameworks: India and USA; Deepfakes and media trust.

1. Introduction

Generative AI is transforming how machines create content by producing text, visuals, and videos at scale.

[1]

Systems

such as ChatGPT, Gemini, and DALL·E now shape a significant share of digital output; approximately 12% of false

online material includes machine-generated media.

[2]

While access to creative tools has expanded, trust in information

has weakened due to growing authenticity risks and ethical concerns.

[3,4]

Ethical oversight frameworks based on

fairness, accountability, transparency, and explainability (FATE) aim to align technological progress with societal

values.

[5]

Organizations such as UNESCO and IEEE advocate people-centric approaches to AI;

[1]

however, country-

level implementations vary. India’s “AI for All” initiative (launched in 2018) emphasizes inclusivity and accessibility,

[6]

whereas the United States’ National Institute of Standards and Technology (NIST) AI Risk Management Framework

1.0 (2023) prioritizes accountability and structured risk management.

[7]

Despite these efforts, public awareness of such

frameworks remains limited, and practical trust in AI systems is correspondingly low. Research examining public

perceptions of responsible AI use in contexts such as synthetic images and voice cloning remains scarce. This study

combines policy analysis with findings from a survey of 40 individuals in India, including students and working

professionals, to explore awareness, trust, and value perceptions related to generative AI. It examines alignment

between policy objectives and public understanding. Unlike prior studies that focus primarily on theoretical ethics,

this work employs empirical evidence to identify gaps between policy intentions and user experience and proposes

approaches to strengthen transparency and responsibility in AI systems across both India and the United States.

2. Literature review

Responsible Artificial Intelligence (AI) rests on fairness, accountability, transparency, and explainability,

[5]

shaping

how emerging technologies are guided ethically. These principles aim to protect individual rights, reduce

discriminatory outcomes, and maintain societal trust in automated systems. Global initiatives, including the

Organisation for Economic Co-operation and Development (OECD) AI Principles (2019), UNESCO’s

Recommendation on the Ethics of Artificial Intelligence (2021), and the European Union’s Artificial Intelligence Act

(2024), promote a human-centric approach to AI governance that prioritizes fundamental rights and social well-being.

However, the rapid growth of generative AI, which can produce realistic synthetic text, images, and speech, poses new

ethical challenges by obscuring content authenticity and source credibility. To address these risks, technical

mechanisms such as the Coalition for Content Provenance and Authenticity (C2PA) framework have been proposed to

embed verifiable provenance metadata into AI-generated content.

[8,9]

2.1 Global ethical frameworks

More than 40 published AI ethics guidelines emphasize FATE principles, namely fairness, accountability, transparency,

and explainability, alongside human rights protection.

[1,3]

Among these, five frameworks are particularly influential at

the global level: the OECD AI Principles (2019), UNESCO’s Recommendation on the Ethics of Artificial Intelligence

(2021), the European Union AI Act (2024), IEEE’s Ethically Aligned Design (2021), and the NIST AI Risk

Management Framework (AI RMF) (2023). While all five frameworks align broadly with FATE principles, they differ

in enforcement strength and operational focus. Some emphasize legally binding compliance mechanisms, whereas

others function primarily as voluntary guidelines or technical standards. Recent governance-focused surveys note that,

despite broad ethical consensus, gaps remain in translating high-level principles into enforceable and measurable

practices, particularly for fast-evolving generative AI systems.

[1,10]

Table 1: Comparative overview of major global responsible AI frameworks.

Framework

Year

Primary Focus

Key Limitation

IEEE Ethically Aligned Design

[1]

2021

Engineering ethics

Complex implementation

NIST AI RMF 1.0

[7]

2023

Technical accountability

Fragmented U.S. adoption

EU AI Act

[10]

2024

Legal compliance tiers

High burden for SMEs

OECD AI Principles

[11]

2019

Voluntary global guidelines

Non-binding; lack enforcement

UNESCO Ethics

Recommendation

[12]

2021

Socio-cultural governance

Limited auditability

These frameworks collectively establish ethical baselines for AI governance, yet their scope, legal authority, and

implementation mechanisms vary considerably across regions.

2.2 National frameworks: India vs. United States

India’s AI for All strategy (2018), together with the NITI Frontier Tech Hub initiative (2024), emphasizes inclusivity,

skill development, and socially beneficial AI deployment.

[6,13]

However, these initiatives function primarily as advisory

frameworks and lack binding regulatory authority. In contrast, the United States’ NIST AI RMF 1.0 (2023) focuses on

structured risk identification, documentation, and accountability across the AI lifecycle.

[7]

Although comprehensive in

technical guidance, its adoption remains voluntary and fragmented across federal agencies and private sectors.

Comparative policy studies indicate that while both countries share core ethical values, their governance models differ

substantially in enforcement mechanisms and institutional coordination.

[1,3]

Recent cross-national analyses further

highlight that generative AI governance remains uneven, with limited standardized mechanisms for content verification

and accountability across jurisdictions.

[4,10]

2.3 Research gap

Despite extensive policy and ethical guideline development, three key gaps persist in the existing literature. First, an

empirical gap remains, as many studies focus predominantly on conceptual or policy-level analysis without measuring

public awareness or perceptions of Responsible AI in practice.

[1,4]

Second, although generative media risks such as

deepfakes and manipulated content are widely acknowledged, research has largely emphasized content creation rather

than user-level validation, detection, and trust mechanisms.

[8,14]

Third, cross-cultural and cross-national comparisons

remain limited, particularly studies that connect national AI governance frameworks with public understanding and

trust outcomes.

[15]

Existing studies focus primarily on policy design rather than public awareness and trust in generative

media, leaving a disconnect between regulatory intent and user experience. This research addresses these gaps by

combining policy analysis with survey-based empirical data to examine awareness, trust, and ethical expectations

surrounding generative AI.

3. Methodology

This study uses both quantitative and qualitative methods, connecting survey responses with official policy documents.

This mixed-methods approach enables comparison across governance systems while exploring individual perceptions

of transparency and ethics.

3.1 Policy comparison

Key policy references include India’s AI for All strategy (2018),

[6]

the NITI Frontier Tech Hub initiative (2024),

[13]

and

the U.S. NIST AI RMF 1.0 (2023).

[7]

Policy documents were qualitatively coded using a deductive framework based

on five Responsible AI dimensions: fairness, accountability, transparency, inclusivity, and enforceability. Each policy

was independently assessed against these dimensions and compared for scope, legal authority, and governance

mechanisms.

Table 2: Comparative overview of selected responsible AI policies.

Policy

Year

Country

Fairness

Accountability

Transparency

Inclusivity

Enforceability

Remarks

AI for All

2018

India

✓

Partial

Advisory

Focuses on

access and skill

development;

non-binding

NITI

Frontier

Tech Hub

2024

India

✓

Advisory

Emphasizes

responsible

deployment and

collaboration;

non-binding

NIST AI

Risk

Manage

ment

Framewo

rk 1.0

2023

USA

✓

Partial

Voluntary

Provides

structured risk

management

guidance;

adoption

fragmented

across agencies

3.2 Survey design

A structured online survey using Google Forms was conducted in September 2025, involving 40 voluntary

respondents. The survey contained 12 items divided into four thematic sections:

1. Awareness & Exposure – familiarity with AI tools and RAI concepts;

2. Reliability and clarity - perceptions of trustworthiness and comprehensibility;

3. Ethical issues - concerns about bias, data misuse, and synthetic media;

4. Policy Focus – opinions on labeling, monitoring, and oversight policies.

Students and early-career professionals were selected due to their high exposure to generative AI tools and their

emerging role as primary adopters. Convenience sampling was employed due to the exploratory nature of the study.

Sample questions included:

• “Have you heard of the term Responsible AI before this survey?” (Yes/No)

• “How important is transparency when using AI tools?” (1–5 scale)

Participation was anonymous, restricted to adults (18+), conducted in accordance with institutional ethical standards.

3.3 Sampling and demographics

Respondents represented both students (85%) and professionals/developers (15%), all based in India. Given the

exploratory sample size (n = 40), the findings are descriptive and intended to identify patterns rather than support

inferential generalizations.

Table 3: Demographic Profile of Respondents (n = 40).

Category

Sub-group

Country

India

100

Primary Role

Students

Professionals/Developers

AI Tool Usage

Daily

Few times/week

22.5

Rarely

2.5

Survey Reliability

Cronbach’s α = 0.84

High consistency

3.4 Analytical procedure

Descriptive statistics and cross-tabulation analyses were conducted using Google Forms outputs and manual

aggregation. Relationships between awareness levels, trust perceptions, and policy preferences were examined through

percentage comparisons and thematic interpretation.

Expanded descriptive cross-tabulation analyses were included to illustrate patterns such as:

• Awareness vs. Trust

• Awareness vs. Support for Labeling

Due to the exploratory sample size (n = 40), inferential statistical tests, such as correlation coefficients or p-values,

were not applied. This approach allows identification of trends and patterns across respondents while linking findings

to the policy frameworks reviewed.

4. Results and analysis

Findings from the public survey (n = 40) are organized around four themes: awareness, trust, policy orientation, and

cross-national perception.

4.1 Awareness of responsible AI and generative tools

The results indicate a high level of engagement with AI tools, with 75% of respondents reporting daily usage, 22.5%

using AI tools a few times per week, and 2.5% reporting infrequent use. Despite widespread usage, awareness of

Responsible AI concepts remained limited. Only 30% of respondents reported a clear understanding of the term, while

32.5% indicated partial familiarity. Notably, 37.5% encountered the concept of Responsible AI for the first time

through this survey. This gap between usage intensity and conceptual understanding highlights the need for improved

awareness initiatives alongside expanding AI adoption.

Fig. 1: Respondents’ frequency of using AI tools (ChatGPT, Gemini, DALL·E, MidJourney) (n = 40).

Fig. 2: Prior awareness of the concept of Responsible AI among respondents (n = 40).

4.2 Trust in AI-generated media and transparency expectations

Trust in AI-generated content was moderate to low among respondents. Only 25% reported high levels of trust, while

a substantial majority expressed uncertainty or skepticism, primarily due to concerns regarding misinformation and

synthetic media manipulation. Transparency emerged as a critical factor influencing trust, with more than half of

respondents rating it as “very important” or higher when using AI systems. Respondents who reported familiarity with

verification mechanisms, such as digital provenance tags or content markers, demonstrated comparatively higher trust

in AI-generated outputs than those without such awareness.

Fig. 3: Respondent perceptions of transparency importance in AI-generated content (n = 40).

4.3 Policy support and content labeling

Strong support was observed for regulatory interventions addressing AI-generated content. A majority of respondents

(65%) favored mandatory labeling of AI-generated material, while 27.5% supported labeling in critical or high-risk

contexts. Only a small minority (2.5%) opposed labeling altogether. Similarly, 62.5% of respondents expressed support

for official oversight mechanisms, indicating broad public approval for governance measures emphasizing

transparency and accountability.

Cross-tab Observations:

• Awareness vs. Trust: Respondents familiar with Responsible AI concepts exhibited higher levels of trust in AI-

generated content compared to respondents without prior awareness.

• Awareness vs. Support for Labeling: Respondents with prior knowledge of Responsible AI principles were more

likely to support mandatory labeling of AI-generated content than those encountering the concept for the first

time.

Fig. 4: Public attitudes toward regulatory oversight of AI systems (n = 40).

4.4 Perceptions of India–U.S. ethical alignment

With respect to cross-national governance, 45% of respondents favored a shared ethical framework between India and

the United States, while 35% preferred initially independent national approaches. The remaining respondents

expressed no definitive preference. Qualitative responses suggested that U.S. frameworks were perceived as structured

and enforcement-oriented, whereas Indian approaches were viewed as inclusive and aspirational. These perceptions

indicate complementary strengths across the two governance models rather than direct opposition.

5. Discussion

Results are assessed in relation to the goals of the research - by contrasting country-level systems, measuring public

understanding and confidence, also sketching approaches for ethical AI use.

5.1 Governance comparison

India's AI for All along with NITI Tech Hub focus on access and basic skills yet enforcement stays limited.

[6,12]

The

U.S. NIST AI RMF 1.0

[7]

provides clearer records along with better responsibility tracking but faces scattered

supervision. That aligns with Kumari,

[15]

whose work highlights ongoing variation worldwide when it comes to actual

enforcement.

Fig. 5: Respondents’ support for mandatory labeling of AI-generated content (n = 40).

Fig. 6: Respondents’ perceptions of ethical alignment between India and United States AI governance frameworks (n = 40).

5.2 Knowing and believing

Even though 75% used AI every day, just 30% grasped RAI ideas - similar to findings by Jobin et al.,

[1]

showing poor

public awareness of ethics. Confidence in generated content stayed weak at 25%, which aligns with transparency

theory.

[14]

Understanding tools like C2PA or digital marks tied to stronger trust; recognition led to greater belief in AI

results.

[9]

5.3 Moving toward clear adoption

Respondents supported content labels (65%) along with government supervision (62.5%), showing clear support for

responsibility measures. A mixed strategy - using tech protections together with public awareness efforts - aligns with

international standards such as human oversight systems, origin tracking data, and openness checks.

[4,8,9]

5.4 New ethical issues

Generative AI boosts creative potential; however, it also increases dangers like fake media and data abuse - key issues

highlighted in the study and often discussed in AI & Society.

[8]

For responsible use moving forward, rules alone aren't

enough - stronger public understanding of tech, moral awareness, and systems that track digital origins must grow at

the same pace.

6. Policy implications and strategic recommendations

6.1 Establishing practical and enforceable guidelines and regulations

Since 65% of respondents supported mandatory labeling of AI-generated content, policymakers in India and the United

States should prioritize the adoption of standardized provenance and content authentication mechanisms, such as the

Coalition for Content Provenance and Authenticity (C2PA) framework.

[8,9]

The strong public preference for labeling

indicates a clear demand for verifiable indicators that distinguish synthetic content from human-generated media,

particularly in high-risk information environments.

6.2 Raising public knowledge while improving skills

Although most respondents reported frequent use of generative AI tools, only 30% demonstrated clear understanding

of Responsible AI concepts, highlighting a significant awareness gap. To address this, both countries should strengthen

AI literacy initiatives by integrating Responsible AI, ethics, and media verification concepts into school and university

curricula.

[6]

These efforts should be supported through partnerships between academic institutions, industry

stakeholders, and public outreach programs to ensure broader societal understanding of ethical AI use.

6.3 Joint India-U.S. governance efforts

Survey findings show that 45% of respondents favored shared ethical oversight between India and the United States,

indicating public openness to cross-national cooperation. Building on this support, a bilateral Responsible AI working

group could facilitate coordination on transparency standards, content verification practices, and watermarking

approaches. Such collaboration would align with UNESCO’s 2021 ethical framework.

[12]

while connecting the

operational guidance of the NIST AI Risk Management Framework with India’s inclusive Responsible AI initiatives.

6.4 Economic or social aspects

While the adoption of transparency and verification mechanisms may introduce short-term compliance and

implementation costs for digital platforms, such measures have the potential to generate long-term societal benefits.

Survey responses indicate strong public support for transparency-oriented governance, suggesting that verified content

labeling and provenance systems can enhance user trust and confidence in digital information environments. Over

time, increased transparency may contribute to more reliable information sharing, greater user engagement, and

sustained public trust in AI-enabled media systems.

7. Limitation

This study is limited by its small sample size (n = 40) and its geographic concentration within urban regions of India.

As a result, the findings may not be generalizable to broader or cross-national populations. Future research should

incorporate larger, more diverse samples, cross-country comparisons, longitudinal designs, and inferential statistical

analyses to validate observed trends, providing stronger evidence for the patterns identified here. Moving ahead,

studies should blend technological, legal, and behavioral insights to build AI governance tools that remain people-

centered-where innovation meets responsibility.

8. Conclusion

This research examined India’s “AI for All” alongside the U.S. NIST AI RMF 1.0, using combined methods to assess

public knowledge and confidence in generative-AI ethics. Results reveal a disconnect between usage and

understanding: while three-quarters engage with AI every day, just 30 percent grasp Responsible AI concepts.

Confidence is weak—nearly two-thirds are unsure or skeptical, but those familiar with openness measures tend to feel

more assured. Support for tagging content is high (65%), while backing for global collaboration stands at 45%, pointing

to a need for clearer, stronger oversight. To close this gap, two paths must align—one enforcing origin rules like C2PA;

another expanding public understanding of AI ethics. In time, ethical AI will not rely on laws alone, but will also grow

from everyday digital habits shaped by shared values.

Conflict of Interest

There is no conflict of interest.

Supporting Information

Not applicable

Use of artificial intelligence (AI)-assisted technology for manuscript preparation

The authors confirm that there was no use of artificial intelligence (AI)-assisted technology for assisting in the writing

or editing of the manuscript and no images were manipulated using AI.

References

[1] A. Jobin, M. Ienca, E. Vayena, The global landscape of AI ethics guidelines, Nature Machine Intelligence, 2019, 1,

389–399, doi: 10.1038/s42256-019-0088-2.

[2] PwC, Global AI Adoption and Risk Report, PwC, 2025.

[3] C. Cath, Governing artificial intelligence: Ethical, legal, and technical challenges, Philosophical Transactions of

the Royal Society A: Mathematical, Physical and Engineering Sciences, 2018, 376, 20180080, doi:

10.1098/rsta.2018.0080.

[4] M. K. Pathan, A. Shah, Ethical governance of generative AI: A systematic review, Journal of Ethics in Emerging

Technologies, 2024, 4(1).

[5] P. Siddhapura, V. Patel, Developing ethical AI frameworks: a comparative analysis of global standards and

practices, TIJER – International Research Journal, 2024, 11, a989–a996.

[6] NITI Aayog, National Strategy for Artificial Intelligence – AI for All, Government of India, 2018.

[7] National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework (AI RMF

1.0), U.S. Department of Commerce, 2023.

[8] S. Lucas, R. M. Heinitz, S. J. Becker, J. E. Charton, Developing a framework for addressing ethical challenges in

generative AI, Journal of Information Technology Case and Application Research, 2025, 1–15, doi:

10.1080/15228053.2025.2558443.

[9] Coalition for Content Provenance and Authenticity (C2PA), C2PA Technical Specification Version 2.0, C2PA, 2024.

[10] European Union, Artificial Intelligence Act (AI Act), Regulation (EU) 2024/1689, European Union, 2024.

[11] OECD, Recommendation on Artificial Intelligence (AI Principles), OECD Publishing, 2019.

[12] UNESCO, Recommendation on the Ethics of Artificial Intelligence, UNESCO, 2021.

[13] NITI Aayog, NITI Frontier Tech Hub: AI for Viksit Bharat, Government of India, 2024.

[14] P. Radanliev, AI Ethics: Integrating Transparency, Fairness, and Privacy in AI Development. Applied Artificial

Intelligence, 2025, 39, doi: 10.1080/08839514.2025.2463722.

[15] P. Kumari, Legal frameworks for AI regulation: a comparative study, Advances in Consumer Research, 2025, 2,

216-224.

Publisher Note: The views, statements, and data in all publications solely belong to the authors and contributors. GR

Scholastic is not responsible for any injury resulting from the ideas, methods, or products mentioned. GR Scholastic

remains neutral regarding jurisdictional claims in published maps and institutional affiliations.

Open Access

This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which

permits the non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long

as appropriate credit to the original author(s) and the source is given by providing a link to the Creative Commons

License and changes need to be indicated if there are any. The images or other third-party material in this article are

included in the article's Creative Commons License, unless indicated otherwise in a credit line to the material. If

material is not included in the article's Creative Commons License and your intended use is not permitted by statutory

regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view

a copy of this License, visit: https://creativecommons.org/licenses/by-nc/4.0/