self-regulation

basicwidespread

Self-regulation means that AI-developing companies set their own rules for their operations and voluntarily commit to following them, without being forced by law.

In plain terms

Imagine a group of friends or sports club members writing their own rules: everyone arrives on time, we return shared equipment in order. These are not prescribed by law, yet they agree because it is better for everyone. Self-regulation is the same in the AI world: companies or an entire industry create their own rules before (or instead of) the state doing so.

In practice, this means ethical guidelines, codes of conduct, voluntary commitments and internal safety policies. For example, a company may promise to test its new model with external experts before release, or to label AI-generated images.

Its advantage is speed and expertise: technology changes in months, legislation takes years, and developers know their own systems best. Its weakness is that if someone violates it, there is often no authority to take action, and the rule is written by those with a vested interest.

How it works

A company or industry association recognizes a risk (for example, misleading content, dangerous use, or agents acting too independently) and decides to act itself.
They document commitments: principles, codes of conduct, or detailed internal regulations on what is tested, what is prohibited, and when development is halted.
The commitment is made public, often signed by multiple companies together, so reputation and competitor scrutiny become the main disciplining force.
Compliance is checked by the company itself; in better cases, an independent auditor or external researcher also reviews it, and results are published.
Over time, established voluntary rules often become standards or laws; signing a code of conduct can be an accepted way to demonstrate compliance with certain regulations.

An everyday example

You open a chatbot and read on the provider's website that, 'under our responsible AI principles,' every new model is tested by independent experts before release. This is typical self-regulation: a company's own promise, not required by law in most places, and difficult for you to verify yourself.

Who is affected

everyoneThe security of AI services used daily often depends on companies' voluntary commitments, not just laws.

entrepreneur, executiveIf you introduce AI at your company, it is advisable to write your own internal policy and assess the commitments of your suppliers.

developerCodes of conduct and internal guidelines directly determine what you can build, how you test, and what you must document. → Code

lawyer, citizenYou must know that if you break a voluntary pledge, you have far fewer legal tools than in the case of a law violation. → Law

A common misconception

MythIf a company has signed a code of ethics, it is as much a guarantee as a law.

In realityA voluntary commitment can be modified or withdrawn by the company at any time, and there is usually no fine for non-compliance. It is not entirely weightless: if a company publicly promises something and fails to keep it, this may be considered misleading consumers. It truly gains weight only when backed by independent verification or legislation.

Limits and risks

There is a conflict of interest: the same company writes the rules that govern the product from which it earns money.
In fierce competition, commitments are easily loosened because whoever slows down for safety may fall behind others.
Beautiful principles sometimes serve only as window dressing (called ethics washing in English), so do not look at the promise but at who is doing the oversight.
It applies only to those who join it; actors that stay out or act in bad faith are not bound by anything.

Similar, but not the same

government regulation – Legislation is written by lawmakers, enforced by authorities, and carries fines for violations, whereas self-regulation is voluntary and relies on internal oversight.
co-regulation – There, the state sets the framework and goals, while the industry develops the detailed rules under regulatory supervision.
compliance – Compliance primarily refers to adhering to externally imposed rules (laws, standards), whereas in self-regulation, the rule itself is created by the company or the industry.

Related concepts

Broader

AI regulation, AI governance, puha jog

Narrower

code of conduct, voluntary commitment, Responsible scaling policy, ethical guidelines

On the edge

co-regulation, AI ethics, EU AI-rendelet, audit, compliance

A brief history

Self-regulation is a long-standing practice in advertising, journalism, and finance, and emerged in the AI industry in the second half of the 2010s: the Partnership on AI was formed in 2016, and the Asilomar AI Principles were drawn up in 2017. In July 2023, leading US AI companies made voluntary safety commitments at the White House. The EU's AI Act, adopted in 2024, explicitly builds on voluntary codes: in July 2025, the Code of Practice for providers of general-purpose AI models was published, and most of the major developers signed it.

Where things stand

At the end of 2026, almost every major AI developer has a public safety or ethics policy, and voluntary codes are also part of EU regulation. In the EU, the emphasis is shifting from pure self-regulation toward co-regulatory solutions that mix in mandatory rules. In the United States, voluntary commitments remain the main factor at the federal level, although some states, such as California, already require by law that safety policies be published.

Status: 2026 November

In the news

Appears in 8 Agent Age stories; the curve shows weekly occurrence in the news magazine.

New Hungarian ad self-regulation recommendation: AI use in advertising must be labelledOct 9, 2026 · Marketing
Trump signs voluntary AI safety pact with industry and names Jay Clayton AI czarOct 5, 2026 · Agent Age
Trump administration has leading tech firms sign voluntary AI safety accordSep 28, 2026 · Agent Age
Trump calls AI self-regulation important; no consensus reached at White HouseSep 28, 2026 · Jog
Governors, attorneys general, Bill Gates and House bills push for federal AI regulationSep 21, 2026 · Jog
Spain drafts AI rules as prime minister says self-regulation is not enoughSep 21, 2026 · Jog
Leading AI labs propose slowdown and shared safety rulesSep 14, 2026 · Agent Age
Microsoft drafts code of conduct on human control of AI and plans to triple computing capacitySep 10, 2026 · Agent Age

Also known as

self-regulation · industry self-regulation · self-regulation · industry self-regulation · voluntary commitments · voluntary commitments · code of conduct · soft law

Sources

OECD AI Principles

This entry was machine-edited and has undergone machine expert review; the “In the news” section updates automatically from Agent Age stories.