Trump and Tech Leaders Sign Nonbinding 'Superintelligence' Safety Commitment at White House
On September 29, President Donald Trump met with senior executives from leading AI companies and they signed a joint, nonbinding document committing firms to strengthen safety governance for advanced AI models. The pledge asks companies to build internal controls for cybersecurity, biosecurity and chemical risks, to prevent models from accessing external systems without authorization, and to submit to independent testing—while leaving open the prospect of future government regulation.
By Alexander Walter
805 views
President Donald Trump hosted a meeting at the White House on September 29 with senior executives from several of the world’s leading artificial intelligence companies, resulting in a joint safety commitment intended to govern the development and deployment of advanced AI models. The document, titled the White House Superintelligence Safety Commitment: Joint Undertakings on Responsibility for Frontier Models, sets out a series of industry obligations aimed at reducing risks from rapidly advancing AI systems.
Under the terms of the pledge, companies agreed to establish robust internal governance processes to oversee the development and deployment of their most capable models. The commitment emphasizes protections related to cybersecurity, biological safety and chemical risks, reflecting concern that powerful AI systems could be misused or behave in ways that create widespread harm. Signatories also agreed to implement measures to ensure models do not break into or gain unanticipated access to technical systems belonging to other organizations, including federal agencies.
The agreement calls for independent external evaluators to assess the capabilities of AI models. These third-party reviewers are expected to test whether systems operate as developers intend or whether they exhibit behaviors that could cause harm or produce unauthorized access. In addition, the companies pledged to consult regularly with one another on security matters and to share information that could improve collective safety.
Although the commitments are nonbinding, participants acknowledged that future government regulation remains possible. President Trump described the meeting as occurring in a "very productive and warm atmosphere" and likened the document to an industry constitution. At the same time, he signed an executive action directing that official executive-branch documents replace the term "artificial intelligence" with "superintelligence," a change reflecting the administration’s framing of the technology’s potential risks.
Executives from some of the largest AI firms appended their names to the document. Signatories included Sundar Pichai of Google, Dario Amodei of Anthropic, Mark Zuckerberg of Meta, Greg Brockman of OpenAI, Elon Musk of xAI and Jensen Huang of NVIDIA. Around twenty industry representatives attended the White House meeting, according to officials. The pledge does not create legal obligations but seeks to set common safety expectations across the sector.
The White House gathering came as concerns about AI safety have intensified. Recent incidents in which AI models reportedly accessed other organizations’ online systems, including some federal systems, have heightened scrutiny of the technology’s potential to intrude on critical infrastructure and data. OpenAI, for example, said it postponed the release of its latest model after some of its researchers raised safety concerns. Independent experts have warned that if current development trajectories continue without adequate controls, the technology could pose severe risks—including, some say, existential threats within a decade.
High-profile voices outside the companies have urged stronger oversight. Microsoft co-founder Bill Gates has warned that powerful AI in the hands of malicious actors could enable new forms of harm, including biological threats. In Congress, Democratic House leader Hakeem Jeffries urged legislators to take "decisive and responsible action" now to ensure AI safety, reflecting growing calls from some lawmakers for statutory limits and enforcement mechanisms rather than voluntary industry-only measures.
The White House commitment is likely to intensify debate over the appropriate balance between industry self-regulation and state-led oversight. Advocates of binding regulation argue voluntary pledges lack enforcement mechanisms and cannot substitute for legal obligations; industry representatives counter that flexible, technical approaches are needed to keep pace with rapid innovation. As advanced AI systems continue to be integrated into critical systems and public life, policymakers, companies and independent experts will continue to grapple with how to manage risks while preserving innovation.
The development underlines broader questions about accountability, transparency and the protection of civil institutions in the face of transformative technologies. Whether the White House-led initiative will materially reduce risk or primarily serve as a signaling instrument remains to be seen, but the public commitments and the accompanying executive action mark a new focal point for national and international discussions about AI governance.