U.S.: Trump backs self-regulation of the AI industry

WASHINGTON, September 30 /ATSH-DPA/ – U.S. President Donald Trump has backed the idea that artificial intelligence companies should oversee and control the development of the technology themselves, instead of new binding rules being imposed by the government.

President Trump met at the White House with leaders of major technology and AI companies, among them Dario Amodei of Anthropic, Jeff Bezos, Elon Musk, Mark Zuckerberg, Sundar Pichai of Google and Jensen Huang of Nvidia.

For OpenAI, the agreement was signed by Greg Brockman, as Sam Altman was at the company’s annual developers’ conference in San Francisco.

After the meeting, President Trump said the companies had agreed on voluntary oversight standards.

He described the approach as a system in which the companies would “police” one another.

The agreement is not legally binding; President Trump has called it “morally binding”.

The signed document provides for strong internal controls to identify and manage the risks associated with AI, including cyber threats, the use of the technology in connection with biological and chemical weapons, and cases in which models might try to enter technical systems without authorisation.

An important element is also the use of independent external auditors.

The companies have also committed to create independent committees at board level to review the reports of internal and external controls.

The agreement includes four levels of control: internal monitoring during the development and training of models, an internal team that checks the functioning of these mechanisms, external auditing, and oversight by an independent board structure.

Nevertheless, the document leaves open the possibility that these measures could be turned into laws or regulations in the future, if that is considered necessary.

For the moment, the agreement is voluntary and does not define a clear state punishment mechanism.

The meeting takes place at a time when several leading AI laboratories have warned of new risks.

According to recent reports, incidents linked to AI agents that have carried out or attempted unauthorised actions in computer systems have increased concerns about the safety of increasingly autonomous models.

Dario Amodei has reiterated that AI can bring very great benefits, but also considerable risks, stressing that the most suitable way to manage them remains part of the discussion.

Meanwhile, Trump has argued that the USA must continue the rapid development of AI to maintain its technological advantage over China.

So, the essence of the agreement: the big AI companies accept to put in place more internal controls and external audits, while the U.S. administration, for the moment, is favouring voluntary self-regulation instead of a new federal regulatory system.