Eight US States Sue OpenAI: Gap Between Child Safety Claims and Terms

01What Was Announced

On October 14, 2026, attorneys general from eight US states filed a lawsuit against OpenAI concerning ChatGPT's safety. According to CNBC's reporting, the central claim is "it presented itself as safe, but reality did not match."

Prior to this, on June 1, 2026, Florida filed the nation's first state-led lawsuit against OpenAI, naming both the company and CEO Sam Altman personally as defendants. A feature of the developments over these four months is that state-level lawsuits came first, and this time multiple states have aligned to contest similar issues together.

OpenAI was not without a response. On August 18, 2026, it launched ChatGPT for Teens, strengthening automatic detection of users under 18 and parental control features. This was announced on OpenAI's official blog. However, the lawsuit continues even after this safety measure was announced. It is thought that the very act of announcing these measures made it easier to contest the gap between safety claims and actual practice as an issue.

02Pricing and Scope of Service

ChatGPT for Teens is an existing mechanism that launched on August 18, 2026, applying age-appropriate restrictions and parental control features to accounts identified as belonging to users under 18. This is stated on OpenAI's official blog.

What needs to be verified is not differences between pricing plans themselves, but how the terms of service for the plan your company has contracted align with the safety claims now at issue. In the news coverage as well, what is cited as the point of contention is the gap between claims and actual operational practice, not a change in contract terms. It is practically useful to directly check the body of the terms of service to see whether this same gap might apply to your own contract.

03Where This Matters for Your Operations

This lawsuit itself does not change how your operations are run. However, the precedent that a gap between a vendor's safety claims and actual practice can become a legal risk is an issue worth referencing regardless of whether generative AI is used in accounting, HR, sales, or planning departments.

If accounting uses ChatGPT to read source documents or draft financial materials, settings related to use for training and whether minors are using the service are not the relevant issues. What matters is whether the safety and accuracy explanations published by the vendor match the assumptions written in your own internal guidelines. The same structure can be used to check HR inquiry handling, sales proposal preparation, and planning department research work.

Although unconfirmed, this lawsuit may prompt a review of how safety claims are stated in enterprise contracts. We believe it is worth establishing an internal process to check for differences in terms of service at the time of contract renewal.

This kind of review can also be an opportunity to take stock of workflows already in place.Building Next-Generation Internal Workflows with Generative AIdiscusses identifying cross-departmental AI processes, which makes it easier to immediately pinpoint the scope of impact when a terms-of-service risk like this one is discovered. For companies expanding AI use through frontline-led initiatives, this is also a scene where the balance discussed inThree Conditions for Successful Frontline-Led AXbetween "frontline discretion" and "company-wide governance" is being tested.

04What to Verify Regarding Governance and Data Handling

There are four points to verify. The first is settings for use in training, i.e., whether the default for your enterprise plan matches your company's configuration. The second is where and how long data is stored, i.e., whether this is clearly stated in the contract or privacy policy. The third is usage authorization, i.e., whether it is defined which departments and roles may use AI for which tasks. The fourth is audit records, i.e., whether it is possible to later trace who entered what and when.

These four points are basic items that should be checked the moment generative AI is introduced into operations, regardless of their direct relevance to this lawsuit. Prompted by this lawsuit coverage, we recommend cross-checking whether the wording in internal regulations matches actual operational practice.

For how to build governance in the accounting domain,Points to Organize Around Governance Before Using AI in AccountingandResolving Key-Person Dependency in Accounting: A Roadmap for Strengthening Governance and Introducing AI Agentsorganize concrete issues. For governance in HR, AI-Only Termination Decisions and HR Governance: Issues Japanese Companies Should Address Early in HR AIcovers state-level regulatory trends on AI-driven termination decisions, addressing the division of judgment between people and AI in HR.

• Whether settings for use in training match your company's contracted plan
• Whether data storage location and retention period are clearly stated in the contract
• Whether usage authorization is defined by department and role
• Whether audit records of input content and output results are retained

05The Next Step

What you should do this week is cross-check the terms of service for the generative AI services your company has contracted against your internal guidelines. Print out the safety and data handling explanations stated on the vendor's official site and read them side by side with the corresponding provisions in your internal guidelines.

If you find places where the wording does not match during this review, we recommend sharing those points with your administrative and IT departments and establishing a plan to correct them by the next contract renewal. The overall design of the process for internally adopting generative AI is covered inBuilding Next-Generation Internal Workflows with Generative AI. Regulatory trends concerning overseas AI vendors are expected to continue, and as discussed inAI Safety and Management Decisions: Issues That Adopting Companies Should Watch, the relationship between safety and management decision-making is drawing attention. We recommend not treating the review of contract terms as a one-time task, but making it a regular practice aligned with these ongoing developments.