Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Repair Windows errors before they cause bigger problems3Scan for outdated or missing drivers - takes under a minuteA voluntary AI safety commitment is a public organisational pledge to take specified steps on safety, security, transparency or reporting. The White House’s September 2023 commitments described testing and red-teaming, sharing safety information, protecting unreleased model weights, inviting third-party vulnerability reports, and developing ways to identify some AI-generated audio and visual content. The pledge describes actions companies agreed to take; it is not a universal legal checklist or proof that a company carried them out.
What did AI companies promise to do?
The White House document, Voluntary AI Commitments, presents a set of practices for participating companies. It says companies making the commitments recognize the importance of information sharing, common standards, and red-teaming best practices. Its commitments cover several parts of how AI systems are developed and made available:
- Testing and red-teaming: Test systems to help identify risks and improve safety and trust. The commitment does not, by itself, specify which systems a particular company tests, which risks it examines, how often it tests, or how it responds to serious findings.
- Information sharing and common standards: Establish or join a forum or other mechanism for sharing information about emerging capabilities, risks, and attempts to circumvent safeguards. The document cites the NIST AI Risk Management Framework as an example that could inform shared practices.
- Cybersecurity for unreleased model weights: Treat unreleased weights as valuable intellectual property, limit access to personnel who need it, use insider-threat detection, and keep the weights in a secure environment.
- Third-party vulnerability reporting: Create a way for outside parties to report issues responsibly, such as a bounty system, contest or prize, or by including AI systems in an existing bug-bounty program.
- Provenance or watermarking: Develop mechanisms such as provenance or watermarking to help identify covered AI-generated audio or visual content, along with tools or APIs that can help determine whether content is AI-generated. This is not a guarantee that all AI-generated content can always be identified.
These are commitment areas and implementation examples, not company-specific test results or product specifications. The document does not establish one standard audit format for demonstrating that a company has met them.
Are voluntary AI safety commitments legally binding?
The word “voluntary” matters: the White House document describes actions companies agreed to take, but the available material does not establish that every commitment is legally enforceable or that falling short automatically triggers a government penalty. Nor does it specify one common remedy for every company that does not fulfill a pledge. The effect of any particular promise depends on its terms and circumstances; the document alone is not enough to make a company-specific legal determination.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
#1 Best Overall
Keep the federal policy timeline separate from private pledges. On January 23, 2025, a later White House executive order revoked Executive Order 14110 and directed a review of policies and actions taken pursuant to it. It told agencies, as appropriate and consistent with law, to suspend, revise, rescind or propose changes to identified agency actions. That establishes a change in federal policy; it does not establish that every company’s separate voluntary pledge was automatically cancelled.
How can you tell whether an AI company is following its safety pledge?
A public pledge gives you a basis for asking for evidence, but the pledge itself does not prove implementation or effectiveness. For a specific company, look for concrete answers in six areas:
- Scope: Which systems, model releases and types of output are covered?
- Specificity: What actions and processes does the company describe, and who is responsible for them?
- Evidence: Does it publish evaluations, audit records or details of how vulnerability reports are handled?
- Transparency: Which risks and results does it disclose, and to whom?
- Security: How does it control access to unreleased weights and address insider threats?
- Accountability: Does it describe progress checks, escalation paths or consequences for unmet commitments?
These questions are a practical way to assess a pledge, not a formal rating standard in the White House document. Without company-specific evidence, you cannot conclude from the pledge alone that a company completed the practices or that they worked.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What the pledge does—and does not—tell you
The September 2023 document sets out areas of intended organisational practice and examples of how companies might implement them. It does not, on its own, show what any named company has completed, establish independent validation, or provide a common enforcement remedy. A sound assessment therefore distinguishes the promise from evidence of performance, and a federal policy change from the status of a private commitment.
Quick Recap
Rank #4
Rank #3
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
