Karma Com Solidarité
Back to the blog

Blog article

Artificial Intelligence: Challenges, Changes and the Search for Accessible, Responsible AI for Non-profits

Artificial intelligence is changing every sector, and it brings real social and democratic risks. Karma Com Solidarité argues that it is time to imagine an AI grounded in solidarity, transparent and sovereign, that serves the common good.

By Zeineb Zhioua

Published on

Updated on

12 min read

Illustration for the article 'Artificial Intelligence: Challenges, Changes and the Search for Accessible, Responsible AI for Non-profits'

Reading

Share

This article reflects the law as it stood in September 2025.

Artificial intelligence is changing every sector, and it brings real social and democratic risks. For Karma Com Solidarité (KCS), a French non-profit that brings volunteers' professional skills to non-profits, businesses and local authorities, it is time to imagine an AI grounded in solidarity, transparent and sovereign, that serves the common good. Here is our analysis.

I. A Technological Revolution with Mixed Effects

Artificial intelligence (AI) is driving a new technological era, with the promise of striking advances in health, finance, education and the environment.

Behind these technical feats, however, lie major social and democratic risks: value concentrated in the hands of a few tech giants, discrimination hidden in algorithms, and wider surveillance of personal data. In response, the European Union (EU) adopted Regulation 2024/1689, known as the AI Act. It sets up a pioneering framework that classifies AI systems by risk level and imposes strict requirements on high-risk applications. In France, the CNIL, the national data protection authority, has backed this framework since 2023 with an "AI plan" and practical guidance sheets, designed to bring the development of AI systems into line with the requirements of the GDPR.

This article looks at the many sides of the AI revolution, from its economic and social effects to its ethical and democratic risks. It proposes an active role for non-profits in promoting an AI that is fair, ethical and regulated.

Illustration for the article 'Artificial Intelligence: Challenges, Changes and the Search for Accessible, Responsible AI for Non-profits'

1. Economic Change and a Shifting Labour Market

Economically, AI is a powerful engine of growth. It automates many repetitive tasks and opens the way to new jobs, while making some professions obsolete. This rapid change in the labour market brings new challenges. It has to be anticipated, people need help through career changes, and suitable training has to be offered, especially to the most vulnerable, so that the social and digital divide does not widen.

AI does more than take over repetitive tasks: it is reshaping the world of work. Some activities are at risk of disappearing, while others are emerging (data steward, AI ethicist, cybersecurity specialist). According to the IMF, AI could affect almost 40% of jobs around the world, replacing some and complementing others. That makes retraining and lifelong learning essential.

For some years now, automation built on sophisticated algorithms has been transforming every sector, from finance to logistics. According to a McKinsey study, generative AI alone could raise labour productivity growth by 0.1 to 0.6 percentage points a year through 2040, depending on how fast it is adopted. Most of these benefits, however, go to large tech companies and highly valued start-ups. That risks opening an economic gap with small and medium-sized enterprises (SMEs), NGOs and less well-resourced regions.

Two people in front of a screen showing a diagram of an artificial intelligence system.

2. Unequal Access and the Digital Divide

Digital inclusion remains a major challenge. In 2023, 15.7% of people aged 15 or over living in France excluding Mayotte were digitally excluded according to Insee's definition, while 30.3% had low digital skills (Insee, 2025). This divide is compounded by differences of age, gender and place. Without targeted measures, such as digital inclusion workshops, equipment loans and local training, AI is likely to push these groups further to the margins.

II. Ethics and the Risk of Discrimination

The rise of AI raises more than economic questions. It also raises major ethical and social issues. Algorithms can reproduce or even amplify discrimination, sometimes invisibly. Automated decisions often lack transparency and can intrude on privacy. All of this raises a basic question about the place of human beings: the development and use of these technologies must guarantee individual autonomy, respect for rights, non-discrimination and fairness.

1. Algorithmic Bias and Discrimination

Algorithms learn from historical data that too often carries society's prejudices, and they reproduce and amplify injustice as a result. A University of Washington study of three large language models used to screen CVs found that they favoured names associated with white people 85% of the time, against 9% for names associated with Black people, and names associated with men 52% of the time, against 11% for names associated with women.

2. Opacity, Accountability and Governance

The idea of a "black box" captures how hard it is to understand how an AI reaches its decisions. Who is responsible when an algorithm gets it wrong and violates someone's rights? The AI Act answers by placing obligations on both providers and deployers of high-risk systems, including risk management, documentation and human oversight.

3. Personal Data and Surveillance

AI systems use huge volumes of personal data for profiling and behavioural prediction. In some "smart cities", algorithmic video surveillance has shown its limits, both on privacy and on discriminatory bias.

III. What Legal Framework for Responsible AI?

Illustration for the article 'Artificial Intelligence: Challenges, Changes and the Search for Accessible, Responsible AI for Non-profits'

1. The European AI Act: A Turning Point

To meet these challenges, the European Union has put in place an unprecedented legal framework: Regulation (EU) 2024/1689, better known as the AI Act, which entered into force on 1 August 2024. This pioneering text regulates artificial intelligence systems according to the level of risk they present.

Uses judged unacceptable are banned. They include social scoring, manipulative or deceptive techniques, untargeted scraping of facial images from the internet or CCTV, and real-time remote biometric identification in publicly accessible spaces for law enforcement, which is allowed only in a few narrowly defined cases. So-called "high-risk" systems, used in particular in health, education, justice, employment and security, are subject to very strict requirements. They must undergo rigorous checks, provide detailed technical documentation, work transparently and be reviewed regularly. Systems that present limited risk are subject mainly to transparency obligations, and minimal-risk systems face no specific obligations. The obligations for high-risk systems apply in stages: the timeline is on the European Commission page listed in the bibliography.

The regulation's main aim is to protect European citizens while encouraging responsible innovation. It seeks to build trust in these technologies and to create effective oversight, and also to preserve Europe's technological sovereignty in the face of large companies from outside Europe. It also provides for regulatory "sandboxes": supervised test environments where new technologies can be tried in real conditions while remaining under legal oversight.

2. The French Framework

In France, as everywhere in the EU, the GDPR completes this framework. Its Article 22 gives everyone the right not to be subject to a decision based solely on automated processing, with the possibility of human intervention and of challenging the decision. The CNIL has published an action plan and recommendations on AI since 2023, with a focus on protecting personal data, limiting data collection and strengthening informed consent. Since 2024, the CNIL has also published practical guidance sheets to help professionals, non-profits included, meet GDPR requirements when developing AI systems.

3. Current Limits and Challenges Ahead

Despite recent progress, some grey areas remain, and they hold back a confident take-up of artificial intelligence, especially in the non-profit sector.

Common criticisms include:

  • Data confidentiality and a higher risk of cyberattacks.

  • Excessive automation, leading to a loss of autonomy and diluted human responsibility.

  • Disinformation amplified by automatically generated content that is hard to verify.

  • The heavy environmental cost of training very large models.

On top of these technical and ethical issues come structural obstacles. Many non-profits struggle to find the money to pay for audits or hire legal advisers, which limits their ability to comply with the new rules.

IV. The Key Role of Non-profits: AI for the Common Good

Illustration for the article 'Artificial Intelligence: Challenges, Changes and the Search for Accessible, Responsible AI for Non-profits'

As artificial intelligence rapidly transforms our societies, non-profits have an essential role in making sure its development serves the public interest. An inclusive approach to AI means four things. It must be accessible to all, even the smallest organisations. It must be ethical: respectful of personal data, transparent in its decisions and free of discriminatory bias. It must be sovereign, favouring technology that is controlled locally. And it must be responsible, backed by suitable training and guidance.

1. Adopting Open-Source AI Tools to Gain Independence

Rather than relying only on proprietary products from the tech giants, non-profits can adopt free and open-source tools that make them more independent.

  • Decidim, an open-source participation platform, lets people shape projects and decisions together in full transparency, while keeping data portable and secure.

  • Open-source Python libraries such as scikit-learn and TensorFlow provide modular building blocks for developing chatbots or analysing volunteer data without depending on commercial APIs.

  • Local hosting in ethical or cooperative data centres gives technological sovereignty.

  • Pooling these resources keeps licence costs down, builds skills and strengthens the trust of the people non-profits serve.

2. Training Volunteers and Vulnerable Groups

Digital inclusion depends on everyone building their skills.

  • Introductory AI workshops that explain machine learning, algorithmic bias and data protection in plain terms.

  • Hands-on sessions on real use cases: a virtual assistant that points people to social services, an analysis of the needs of disabled people, and so on.

  • Data literacy courses run with community spaces ("tiers-lieux", shared local venues such as coworking spaces and makerspaces) to sharpen critical thinking about algorithmic recommendations.

These actions encourage responsible use of the tools, give vulnerable people more control, and make AI available to more people.

3. Acting as Watchdogs and Making Proposals

Non-profits also need to weigh in on public debate, so that AI remains a tool for social justice.

  • Taking part in European consultations and CNIL working groups, to bring what non-profits see on the ground into the discussion and to defend the rights of the most vulnerable.

  • Working with independent civil-society observatories to produce reports on the social effects of technology.

  • Organising public forums and participatory workshops to draft ethical charters or legislative recommendations together.

Through these initiatives, non-profits can help build an AI that is fairer, more transparent and more respectful of freedoms, and that combines technological innovation with human solidarity.

Conclusion: What AI Do We Want for Tomorrow?

Artificial intelligence is a choice about the kind of society we want, as much as a matter of technology and algorithms. Europe has laid solid foundations with the AI Act, the GDPR and various national measures, but putting them into practice remains a challenge for grassroots organisations, especially the smallest. Non-profits are central to this transformation, thanks to their capacity for civic innovation, their local roots and their role as go-betweens. They can adopt open-source tools, pool resources, train people and act as ethical watchdogs in democratic debate.

Building an AI that is responsible, inclusive and frugal takes a collective effort, with citizens, public institutions, businesses and civil society moving forward together. On that condition, AI can become a lever for emancipation, social justice and sustainability, one that enriches human intelligence instead of replacing it.

Bibliography

1. Articles, Reports and Legal Journals

2. Laws and Regulations

By Zeineb Zhioua, legal professional specialising in international and European law

Read next