NORBERT BIEDRZYCKI

ChatGPT and the New Ethics of AI: What Happens When Everyone Has Access to Intelligence?

ChatGPT and the New Ethics of AI: What Happens When Everyone Has Access to Intelligence?

ChatGPT reached an estimated 100 million users in about two months, faster than any consumer application before it. Six months after its launch, the ethical questions it raises look very different from those we discussed only a year ago.

A few episodes from this spring illustrate the shift.

At the end of March, Italy’s data protection authority temporarily blocked ChatGPT, citing concerns about the legal basis for processing personal data and the absence of age verification. Service was restored a month later after OpenAI made changes.

In April, engineers at Samsung pasted confidential source code and internal meeting notes into ChatGPT to help with their work. Samsung responded by restricting the use of generative AI tools on company devices.

This month, a federal judge in New York sanctioned two lawyers who had submitted a court filing citing six cases that did not exist. ChatGPT had invented them, and the lawyers had not checked.

From specialized systems to universal access

Until last year, the ethical debate about AI was largely about systems built and deployed by organizations: credit models, hiring algorithms, facial recognition. The question was how those organizations should behave.

Generative AI changes the unit of analysis. The systems are now in the hands of hundreds of millions of individuals, used for purposes nobody designed, tested or approved. The risks are no longer only about how AI is built. They are about how it is used.

The new risk profile

Confident error. Language models produce fluent, plausible text whether or not it is true. The Avianca case is funny until it happens in a medical, legal or financial context with real consequences.

Data leakage. Employees use public tools with confidential information because it makes them more productive. Most organizations had no policy for this in January.

Bias at scale. Models trained on internet text inherit its biases and reproduce them in millions of interactions.

Misuse. The same capability that writes a marketing email writes a convincing phishing campaign.

Where responsibility sits now

Responsibility is now distributed across at least three parties: the model developer, the organization that deploys or permits the tool, and the individual user. Each has a role. The developer can build safeguards, the organization can set policies and train people, the individual can check before trusting.

For companies, the practical conclusion is urgent. Banning generative AI does not work, because employees will use it on their own devices. Ignoring it does not work, because the risks are real. The only workable path is to provide approved tools, clear rules on data and verification, and training that explains not only how to use the tools but where they fail.

The deeper shift is cultural. For the first time, every employee needs a basic understanding of how a powerful AI system behaves. AI literacy is becoming part of professional competence, much like data protection awareness did after GDPR.

Leadership question: Who is responsible for ensuring that powerful AI tools are used safely when access to them is becoming almost universal?

Read next

  1. Ethical AI: From Principles to Accountability, 2020–2026

  2. The Ethics of AI and Human Autonomy: Are We Giving Away Too Much Control?

    Ethical AI series, written in 2026

  3. The Accountability Gap: Who Is Responsible When AI Makes Decisions Independently?

    Ethical AI series, written in 2026