Ethics
Ethics
Digital safety
International Day of AI: Top AI Trends, EU AI Act Updates, and the Future of Responsible AI
Mary, NexSynaptic Founder
International Day of Artificial Intelligence: Why It Matters Today
The International Day of Artificial Intelligence (Day of AI) is a global initiative celebrated every June by the AI community. Although the United Nations has not yet declared an official “International Day of AI,” artificial intelligence is already part of everyday life from work and learning to healthcare, finance, and public services.
With the EU AI Act becoming legally binding in 2026, this day is an opportunity to take a clear look at what AI does, where it helps, where it creates problems, and how to use it in ways that are beneficial rather than harmful.
Three core areas shape this conversation: understanding artificial intelligence, developing artificial intelligence, and responsibly applying artificial intelligence.
Understanding Artificial Intelligence: From Technology to Societal Impact
Understanding AI today requires knowing what AI does in practice, how it shapes our work, learning, and decisions, and the legal framework it operates within (e.g., the AI Act). AI is “software in the background” part of the infrastructure that influences how we work and learn, how we make decisions and access information, and how banks, hospitals, schools, media, and public institutions function.
To use AI safely, it is essential to understand how models learn, what data they use, where and why they can fail, how bias emerges, and how their performance is evaluated. Without this, AI can produce unfair, discriminatory, or simply incorrect decisions.
Real‑world examples: when misunderstanding becomes costly
Argentinian court and ChatGPT (2025)
A judge reportedly used ChatGPT to help prepare a ruling without disclosing it. The ruling was later overturned because the AI invented legal references (“hallucinated”). The model sounded convincing but mixed accurate and inaccurate information in ways that non‑experts cannot easily detect.Canadian tax authority chatbot (2025)
A national tax chatbot provided incorrect information to thousands of citizens about tax credits and deadlines. The issue was not malicious intent but insufficient verification of what the model actually did before it was released to the public.These cases show that understanding AI is a prerequisite for using it safely.
Societal AI impact: who is affected?
AI affects employment and the labor market, education and skills, healthcare and diagnostics, finance and credit scoring, democratic processes and public discourse, mental health, privacy, and security.
This is why the AI Act places special emphasis on how AI affects fundamental rights.
Real‑world examples of societal impact
Hiring tools and discrimination
Several companies used AI systems to automatically sort CVs. Research showed that some of these systems systematically penalized female candidates or candidates with “non‑typical” career paths because they were trained on historical data where men were more frequently hired for certain roles. This is why hiring tools are classified as high‑risk under the AI Act.
Credit scoring and denied loans
AI credit‑scoring systems in some countries automatically rejected loans for entire groups of people based on proxy variables (e.g., postal code, phone type, spending patterns), which was effectively discriminatory. The AI Act therefore specifically regulates AI that affects access to essential services such as credit.Mental health and chatbots
In 2025, U.S. attorneys general warned major AI companies about chatbots that gave dangerous, “affirming” responses to users with depressive thoughts instead of directing them to professional help. A study from Brown University showed that such systems can increase self‑harm risk among vulnerable users.
Developing Artificial Intelligence: From Models to Sustainable Systems
Developing AI systems today cannot be reduced to “making them work technically.” These systems must be safer, more resilient to errors and attacks, and sustainable and compliant with regulations. Key development trends include security assessments and red teaming (intentional testing of weaknesses), adversarial robustness, reducing hallucinations in generative models, and increasing transparency and explainability. This is important for generative systems that increasingly interact with people.
Privacy‑preserving development is gaining importance: federated learning (data stays local), differential privacy, and encrypted training. It is crucial in sectors healthcare, finance, and neurotechnology, where data is extremely sensitive.
AI development now involves engineers, ethicists and lawyers, psychologists and neuroscientists, security experts, and UX designers. The goal is to build AI systems that are useful, safe, and socially acceptable.
Real‑world examples from development and safety
Rising incidents and declining transparency (2024–2026)
According to the Stanford AI Index, the number of documented AI incidents increased from 233 in 2024 to 362 in 2025, while the average transparency of foundation models dropped from an index score of 58 to 40. This means models are becoming more powerful, but we increasingly know less about how they were trained, what data they use, and what risks they carry.
Privacy and data leaks (2025)
In 2025, AI‑related privacy and security incidents increased by more than 56%, with hundreds of cases in which personal data leaked or was misused through AI systems (e.g., models accidentally “revealing” training data, poorly configured APIs, unprotected fine‑tuning datasets).
Responsible Application of Artificial Intelligence
Responsible AI use in business becomes a key topic for 2026, especially because the AI Act becomes legally binding.
Transparency toward users
Responsible application means users must know when they are interacting with an AI system, how the system makes decisions, what data it uses, and what its limitations are. Transparency is the foundation of trust and an explicit requirement of the AI Act.
Real‑world examples of transparency
City councils and algorithmic reports
Some European city councils already publish “algorithmic transparency reports” explaining which AI systems they use for allocating social services, what data they rely on, what their purposes are, and what limitations they have.
Hiring platforms and decision explanations
Some candidate‑matching platforms now allow applicants to see why they were or were not matched with a particular position — which skills, qualifications, or experiences influenced the decision. This reduces the “black box” effect and increases trust.
Human oversight
AI systems must not operate fully autonomously without control. Responsible application includes human‑in‑the‑loop, human‑on‑the‑loop, clear intervention protocols, and the ability to override or correct decisions.
This is especially important in high‑risk areas such as hiring, credit scoring, healthcare, and justice.
Real‑world examples of human oversight
University admissions
Some admissions offices use AI for preliminary sorting of candidates, but all borderline cases and rejections are reviewed by humans. This ensures that the algorithm does not make final decisions about access to education.
Banks and credit approvals
Some banks configure AI so that it automatically approves a loan only if the model is more than 90% confident. All other cases go to experienced credit analysts who can consider individual circumstances the model cannot see.
Fairness and non‑discrimination
Responsible AI requires unbiased training data, fairness metrics and discrimination tests, measures to mitigate inequality, and impact assessments across different social groups. Without this, AI can easily reproduce and amplify existing inequalities.
Real‑world examples of bias mitigation
Monthly bias audits in online lending
Some fintech companies conduct monthly audits of loan approvals across demographic groups. If the difference in approval rates between protected groups exceeds 5%, automatic retraining is triggered. This is a concrete example of turning fairness principles into operational processes.
Synthetic candidates in recruitment testing
Some developers of hiring software test their algorithms with synthetic candidate profiles of different genders, ethnic backgrounds, and educational paths to detect systemic biases before launching the product.
Security and resilience
Responsible application also includes protection against attacks and misuse, anomaly detection, continuous performance monitoring, incident reporting, and learning from failures. This is crucial for the security and resilience of AI models in the real world enviroments.
According to the data presented in the Stanford AI Index Report 2026, the rising number of documented AI incidents shows that AI is advancing at an extraordinary pace, while security, regulation, education, and public trust are not keeping up.
Real‑world example of incidents and learning
Police reports and amplified racial bias
In some jurisdictions, AI systems used to summarize police reports showed a tendency to amplify racial stereotypes because they learned from historical data in which certain groups were more frequently described negatively. After public pressure and investigations, these systems were withdrawn or revised with stricter human oversight and bias testing.This case shows that security is technical (attacks, hacks) and social.
Compliance with Regulations: The AI Act as a Framework for Responsibility
The AI Act sets clear standards for high‑risk AI systems, foundation models, transparency and user notification, documentation and system monitoring, human oversight, and safety. Responsible application means compliance from day one not retroactive “fixing” when inspections or fines arrive.
From August 2, 2026, most key obligations of the AI Act become legally binding, including rules for high‑risk systems, transparency for chatbots and generated content, and stricter risk‑management requirements. Some EU countries are already preparing early enforcement activities: national authorities are reviewing reports of prohibited practices (e.g., emotion recognition systems in workplaces), and the EU AI Office is monitoring obligations for providers of large foundation models.
International Day of AI: A Reminder of Responsibility
The International Day of AI is a reminder that artificial intelligence is not just a technological tool but a form of social infrastructure. It shapes how we live, work, and make decisions. Understanding, developing, and responsibly applying AI are the three pillars that determine how useful, safe, fair, and aligned with European values AI will be.
Note: The International Day of AI is a regional and community‑driven initiative of the global AI community, not an official United Nations or UNESCO observance.
FAQ About the International Day of AI and the EU AI Act
1. Why is the International Day of Artificial Intelligence celebrated?
The International Day of AI is celebrated to promote understanding, responsible development, and safe use of artificial intelligence across society.
2. Why is 2026 an important year for AI?
In 2026, the EU AI Act becomes legally binding. It is the first comprehensive regulation defining how AI can be developed, deployed, and supervised within the European Union.
3. What changes does the EU AI Act introduce?
The Act requires transparency, strict rules for high‑risk AI systems, human oversight, detailed documentation, continuous monitoring, and bans practices that threaten fundamental rights.
4. How does AI influence everyday decisions?
AI already shapes hiring, credit scoring, healthcare diagnostics, education, public services, and digital security — often invisibly, but with significant impact.
5. What are the biggest risks associated with AI systems?
Key risks include biased data, hallucinations (fabricated information), lack of oversight, unclear algorithmic decisions, and misuse of personal data.
6. What does it mean when an AI system is “high‑risk”?
High‑risk AI systems are those that can affect fundamental rights, access to essential services, or personal safety — such as hiring tools, credit scoring models, healthcare systems, and public administration algorithms.
7. How can organizations ensure responsible AI deployment?
Responsible deployment requires human oversight, clear transparency, bias testing, privacy protection, and continuous performance monitoring.
8. How can companies prepare for the EU AI Act?
Companies should assess the risks of their AI systems, implement documentation, conduct safety and ethics testing, ensure human oversight, and align internal processes before the Act takes effect.
Explore more insights on ethical, transparent, and human‑centered AI in the NexSynaptic Ethics Guide.
Explore related pillars
👉Digital Safety Guide - A framework for privacy and safe digital ecosystems.
👉Neurotechnology Guide - A framework for neural data, BCI technologies, and neuromorphic computing.
Browse all Ethics articles👉 ethics
Read more about AI Ethics & Regulation👇
🔙 Return to the beginning of the journey
Explore more topics:
Ethics • AI Trends • Neurotechnology • Digital Safety • Brain Science • AI Tools • Technology
