Leaders of some of the largest US technology and artificial intelligence companies have signed an industry self-regulation agreement on AI safety, which US President Donald Trump described as an important step towards managing risks associated with increasingly advanced systems.
The framework, however, does not impose legally binding obligations or penalties. At the same time, Trump has formally moved to change US government terminology from “Artificial Intelligence” to “Super Intelligence”.
Voluntary safety agreement
Trump announced the “Joint Commitment On Frontier Responsibilities” following a White House lunch attended by leading figures from the US technology industry.
The president presented the agreement as a framework intended to provide protection against potential risks accompanying the development of new artificial intelligence systems.
“It’s almost like a constitution. The greatest people in the world signed it and I signed it as president. It really is a form of protection,” Trump said.
The framework, however, does not provide for legal obligations, sanctions or direct participation by government regulators.
According to international media reports, including The Guardian, it effectively constitutes a voluntary commitment under which companies undertake to implement their own mechanisms for monitoring and assessing the safety of their products.
From ‘Artificial Intelligence’ to ‘Super Intelligence’
Alongside the agreement, Trump signed an executive order formally changing the terminology used by the federal government for artificial intelligence.
Under the order, federal agencies are instructed to use “Super Intelligence” and “SI” in official documents, public announcements and policy texts, replacing “Artificial Intelligence” and “AI”.
The move follows similar language used by Trump at the UN General Assembly, where he referred to the next stage in the development of artificial intelligence systems.
Four levels of company-led safety checks
The agreement provides for four levels of safety checks and assessments for companies developing artificial intelligence systems.
The first requires internal mechanisms to monitor safety while models are being trained.
The second provides for the creation of an internal team to assess whether those mechanisms are operating effectively.
At the third level, companies commit to allowing external evaluators to independently examine their safety procedures.
The fourth provides for the establishment of independent boards to review the relevant reports.
Despite these four stages, the agreement does not require companies to publish the results of their assessments. Companies are also left to choose their own external auditors and supervisory bodies.
Calls for tighter regulation
The agreement comes amid increased pressure on artificial intelligence companies following incidents that have raised concerns about the safety of their systems.
Lawmakers, researchers and some figures within the technology industry have called for stronger government oversight.
Trump has repeatedly rejected extensive restrictions, arguing that they could slow innovation and weaken the United States in its competition with China.
Musk, Zuckerberg and Bezos attend White House meeting
Among those attending the White House meeting were Elon Musk, Meta CEO Mark Zuckerberg, Amazon founder Jeff Bezos, Google CEO Sundar Pichai, Nvidia founder and CEO Jensen Huang, OpenAI co-founder and president Greg Brockman and Anthropic CEO Dario Amodei.
House Speaker Mike Johnson, who played a leading role in organising the meeting, told CNBC that the United States needed to find the “right balance” between safety and its objective of becoming the global leader in artificial intelligence.
Zuckerberg described the agreement as an important first step towards strengthening public confidence.
“This is a beginning and an agreement that the entire industry could come to,” he said. “It was a historic discussion.”
