Current Region:
Global

Between now and the end of the 119th Congress in January, there is a real chance that the House and Senate child safety bills could be combined and passed as a single package. As they stand, both drafts would set a federal minimum level of protection for children online while still leaving individual states free to go further.  This is perhaps one topic where the incoming Congress might hope to pass legislation that is not vetoed by the President, so potentially the current Republican leadership may decide to get it passed now while they can still shape the outcome.

The two chambers are not yet aligned. On 29 June the House passed H.R. 7757, the Kids Internet and Digital Safety (KIDS) Act, by 267 to 117. On 5 August the Senate Commerce Committee advanced an amended S. 1748, the Kids Online Safety Act (KOSA), by voice vote. On 3rd November, 45 state and territorial attorneys general wrote to congressional leaders urging passage.

We leave policy decisions to politicians but set out here our technical advice on the age assurance aspects of any new legislation whenever it is considered.

What a combined bill should include

From an age assurance perspective, an effective combined bill should:

  1. Restore an accuracy requirement and empower the FTC to set minimum, outcome-based performance standards, informed by recognised standards such as IEEE 2089.1 and ISO/IEC 27566-1, with measurable performance appropriate to the risk and age threshold.
  2. Recognise independent certification and provide a safe harbor for the relevant age assurance obligations only where the complete system, as implemented by the platform, is independently assessed and shown to meet the required outcomes. This must cover the combination of methods, configuration, integration, security, anti-circumvention controls and real-world performance. Certification of an individual component or provider alone must not confer a safe harbor, nor remove the platform's responsibility for other child safety duties.
  3. Adopt one consistent knowledge standard across all titles, backed by FTC guidance explaining how properly implemented and independently assessed age assurance can help demonstrate compliance or benefit from safe harbor.
  4. Remain technology neutral, allowing estimation, verification, inference, reusable digital credentials.
  5. Keep strict privacy safeguards: data minimisation, prompt deletion when data is no longer needed, double-blind and zero-knowledge options and no requirement to use government ID. Age assurance should establish age eligibility without unnecessarily revealing identity.
  6. Act as a floor, not a ceiling, with express wording that saves state age assurance laws and keeps child safety out of any wider pre-emption of state AI laws.
  7. Extend the Commerce Department/FTC/FCC study to cover all approaches to age assurance and interoperability, not only device-based methods.

The current House version

The KIDS Act bundles more than a dozen proposals. For age assurance, the key provisions are:

  • A direct mandate for adult content. Title I, the SCREEN Act, applies to platforms where more than a third of the material is sexual material harmful to minors. From a year after enactment, these platforms must use commercially available "technology verification measures" to identify users under 18 and stop them reaching that material.
  • A somewhat equivocal performance standard. A technology verification measure is defined by whether it determines that a user is "more likely than not" a minor. This describes a decision threshold, not a measured system-wide accuracy rate or a minimum assurance level. The introduced bill required measures to be "reasonably designed to ensure accuracy", but that phrase was dropped from the final text. The resulting absence of objectively measurable minimum performance requirements risks inconsistent implementation.  It also implies that the test is applied to minors, when, as with entry to a bar, age has to be proven by adults.
  • Sensible design rules. Self-declaration is not enough. Platforms must take reasonable measures against circumvention, keep and use verification data only as strictly necessary, publish how they verify age and maintain reasonable data security. They may choose any method that meets these rules, may use third parties (while staying liable) and cannot be required to collect government ID.
  • A limited role for the FTC. The Federal Trade Commission must consult experts, including age assurance providers, on "standards and metrics". However, the SCREEN Act gives it no power to make rules, so any standards it develops would be advisory.
  • No mandate for social media. The KOSA subtitle applies its duties only to users a platform "knows or should have known" are under 17. It states that, unlike SCREEN,  it does not require age gating or age verification. COPPA 2.0 contains the same disclaimer and adopts an "actual knowledge or should have known" standard for children under 14 and teens aged 14 to 17. The general provisions add that nothing in the Act requires collecting age information not already collected in the normal course of business.
  • AI chatbots and games. The SAFE BOTs Act and the Safer GAMING Act also use "know or should have known" but do not contain that age verification disclaimer. The SPY Kids Act and the data broker provisions keep the older "actual knowledge or willful disregard" standard.
  • Best practice. A Commerce Department Kids Internet Safety Partnership would publish a playbook covering age verification, assurance and estimation techniques.

The current Senate version

S. 1748 takes a different route. This comparison concerns the Senate's KOSA text, not every child safety proposal considered by the Senate. On 5 August, the Senate Commerce Committee also voted 15–13 on the separate SCREEN Act (S. 737), although a further vote was required under committee attendance rules. Its possible inclusion in any wider package therefore remains a separate question:

No Age Assurance mandate in S. 1748. The Senate KOSA text has no adult content title and states that it does not require age verification or age gating. This should not be confused with the separate Senate SCREEN Act proposal.

  • A constructive knowledge standard with guidance. Duties apply where a platform has actual knowledge of a minor, or knowledge "fairly implied on the basis of objective circumstances". The FTC must issue guidance on applying this standard, including best practices and examples.
  • A broad duty of care. Platforms must exercise reasonable care in designing features to prevent and mitigate specified harms. The duty covers online video games, messaging apps and video streaming services as well as social media. Reliable age assurance could be an important means of demonstrating compliance, enabling platforms to apply appropriate protections to minors without imposing unnecessary restrictions on adults. For this reason, some commentators argue it will drive adoption of age assurance. It would not, on its own, discharge the broader duty of care.
  • A study of device-based verification. Section 107 directs the Commerce Department, with the FCC and FTC, to study whether age can feasibly be verified at the device or operating system level. The study covers accuracy, accessibility, data minimisation, privacy, security and competition.

Pre-emption

Both versions seek to preserve state laws offering greater protection to minors while pre-empting conflicting requirements. The precise boundary between additional protection and a direct conflict may still require judicial interpretation. The House moved a long way to get there. The introduced KIDS Act would have barred states from maintaining any law requiring age verification for harmful material. That clause was replaced in the June bipartisan deal, and the adult content title now also expressly preserves state privacy and data security laws. The apparent intention is that many existing state age assurance laws for adult content, social media and AI chatbots would survive either version, subject to the wording and operation of each particular state law:

  • Adult content. State laws with higher assurance levels, lower content thresholds, private rights of action or rules based on a user's actual location may qualify as greater protection. Whether particular requirements survive would depend on any direct conflict with federal law. Under the House version, age assurance would also become mandatory in states that have no law of their own.
  • Social media. The disclaimers only say how the federal text should be read. They create no federal right to avoid verification, so they would not necessarily conflict with state laws that require it. The outcome would depend on the specific state provision.
  • AI chatbots. The House chatbot title uses the same pre-emption formula, so stronger state companion chatbot laws may survive, although specific conflicts would need to be assessed. The greater risk lies outside these bills: the reported deal between Senator Blackburn and the White House would link child safety to federal pre-emption of some state AI laws.
  • App stores. Neither text addresses app store age verification directly, but a wider package incorporating the App Store Accountability Act would change that. Any app store approach should preserve competition, interoperability and the ability of independent providers to supply age assurance..

Conclusion

The central question for Congress is not simply whether platforms should verify age. It is how they can reliably distinguish children from adults, apply proportionate protections and demonstrate that their systems work. Legislation should establish clear outcomes rather than prescribe a single technology. Independent assessment of complete, deployed systems, appropriate accountability and interoperability between providers would support effective, privacy-preserving solutions. Ultimate liability must remain with those operating risky services. 

The lame duck window is short, but a clear, enforceable federal floor that leaves states free to go further could provide a durable foundation for protecting children online.