Online safety regulation tracker
| SevereBuild or respond: change the product, or stand up a response. | 7High | 16Critical | ||
|---|---|---|---|---|
| MajorDocument or brief: publish, file, or brief your team. | 2Moderate | 1High | 7High | |
| ModerateAssess: check what you already have. | 3Low | 7Moderate | 8Moderate | |
| MinorWatch: nothing to do yet. | 2Low | 3Low | 3Low | |
| Severity ↑ Likelihood → | UnlikelyDoesn't apply to you | PossibleNot final yet, or applies to other services | LikelyApplies to you from a set date | CertainApplies to you now |
LowModerateHighCritical
Moderate and possible
3 shown
The deepfake harms that arrived were not the ones detectors were built for
Published researchGlobalResearch2026AssessLow if you run user-to-user, marketplace and messaging and other consumer services anywhere, at any size.
LowModerate and possibleNo date setNew York's RAISE Act requires frontier AI incident reports within 72 hours from 2027
MultiStateUSDeadline27 Mar 2026AssessLow if you run generative AI with users in the US, at any size.
LowModerate and possibleNo date setThe FTC's study of AI companion chatbots and children is still open, with no findings yet
Federal Trade CommissionUSOperation11 Sep 2025AssessLow if you run generative AI, user-to-user and messaging with users in the US, at any size.
LowModerate and possibleNo date set
Not legal advice. Read the source.