Tomorrow, August 12, 2026, over 100 new laws take effect in Colorado. One of these will directly impact everyone using Character.AI, Replika, or JanitorAI for companionship or roleplay—HB 1263, a new law specifically regulating “conversational AI services.” This is no longer an isolated case: from California’s SB 243, the federal GUARD Act, to the 11-state overview we compiled last time, the number of US states with specific AI companion laws has grown from 11 to 14 in just three months, with one state now resorting to criminal law for violations. This article explains what the new Colorado law actually mandates, the key points of the 3 newly added states, and whether this wave of regulation affects you if you use platforms like WeavAI.
Taking Effect Tomorrow: What does Colorado HB 1263 mandate?
The Colorado General Assembly adjourned on May 13, 2026. According to state custom, bills without specific “safety clauses” automatically take effect on a fixed date after adjournment—which is tomorrow, August 12, 2026. HB 1263, led by Democratic Representatives Sean Camacho and Javier Mabrey, is officially titled “Regulations for Conversational Artificial Intelligence Service Operators” and applies to all AI services “open to the general public whose primary function is to simulate human conversational interaction.”
Breaking down the bill, the logic is consistent with regulations we’ve tracked in other states. Key points include: AI identity disclosure; prohibiting the use of points or reward mechanisms to induce minors to interact continuously; implementing reasonable measures for minor users to avoid generating sexually suggestive content or content simulating emotional dependency; establishing response protocols for self-harm/suicidal ideation; and prohibiting AI from implying it is equivalent to a licensed therapist.
What makes this Colorado bill stand out is the penalty—a fine of up to $20,000 per violation, with no total cap, enforced under the Colorado Consumer Protection Act. This means that if a platform has systemic issues, fines can stack infinitely, representing one of the heaviest financial pressures among similar state regulations currently.
This is the fifth piece of the puzzle this year, not breaking news
If you recall, this is our fifth follow-up report on US AI companion regulation this year. From California’s SB 243 taking effect last year, to the federal GUARD Act passing the Senate Judiciary Committee 22-0, to the 11-state overview we compiled in July, this timeline has been steadily advancing. Colorado just happens to be at a particularly visible point: taking effect tomorrow, with one of the highest penalty amounts known to date.
3 new states in just three weeks, now at 14
When we last compiled the 11-state list, Colorado was already included, but it hadn’t taken effect yet. The real update is that the list itself has expanded—Hawaii, South Carolina, and Wyoming have joined, bringing the total from 11 to 14.
Hawaii’s SB 3001 was signed into law on July 14, requiring AI operators to disclose information to account holders and establish protocols for content related to suicidal ideation. South Carolina’s H 3431 focuses on “safety by design” obligations for minors, requiring operators to proactively reduce risks of addiction, psychological harm, and privacy infringement. Wyoming’s HB 102 is an entirely different level of regulation, worth looking at separately.
Wyoming is unique: The only state to invoke criminal law
For the other 13 states, regardless of how heavy the fines are, they are essentially “civil” regulations—if violated, the state government or users can seek damages or issue fines, but no one from the company goes to jail. Wyoming’s HB 102 is different. It classifies “developing or distributing AI systems intentionally designed to generate deepfake sexual content for children, or intentionally designed to encourage self-harm” as a criminal felony, punishable by up to 10 years in prison, plus a $10,000 fine.
The background for such heavy legislation is statistics cited by the Wyoming legislature during deliberations: approximately 1.2 million children are affected by deepfake image generation technology annually, with the related industry estimated at $1.5 billion. This isn’t a declaration of war by Wyoming against AI companion platforms as a whole; the bill targets the two most malicious abuse scenarios: deepfake pornography and encouraging self-harm. However, raising the sentencing to the criminal level is unique among the 14 states and provides the first concrete legal basis for the idea that “AI companies may face real criminal liability.”
What do these regulations mean for you, as a user of Character.AI or Replika?
Seeing the word “regulation” makes it easy to intuitively think, “Are they banning AI companions?” But looking closely at the provisions in these 14 states, the core spirit is highly consistent: requiring operators to honestly disclose AI identity, not to use manipulative tactics to get minors addicted, to have referral mechanisms when encountering self-harm tendencies, and not to let AI impersonate professional psychological counseling. For the vast majority of adult users who treat AI roleplay as a hobby or a way to relieve stress, these regulations don’t change “whether you can use it,” but rather “you will see more safety prompts and clearer privacy setting options when using it.”
The most directly affected are platforms that have historically been lax with minor protection or whose business models rely heavily on “emotional dependency” design—which is why platforms from Character.AI to Replika have been adjusting age verification, conversation limits, and disclosure mechanisms over the past year. To some extent, these regulations are raising the uneven industry standards to a common baseline.
WeavAI’s Perspective: An all-ages platform, no homework needed
WeavAI has been positioned from the start as an all-ages interactive fiction and roleplay platform. It does not generate pornographic or NSFW content, nor does it rely on manipulative designs that “make you addicted to AI characters” to generate revenue—the payment mechanism is built on a positive cycle of 30% credit rebates for creators and +500 credits for daily check-ins, rather than emotional attachment. The problems these 14-state laws aim to eradicate (misleading minors, sexually suggestive content, encouraging emotional dependency, lack of self-harm response mechanisms) were never part of the WeavAI business model.
This is why we haven’t written about this news as a “crisis,” but rather as a continuous follow-up—stricter regulation is actually a relatively favorable long-term signal for platforms that are already on a safe path.
What’s next?
The Colorado legislature’s session for this year hasn’t ended yet, and more states are expected to follow suit before the end of the year. We will continue to track this series—if you are a long-term user of Character.AI, Replika, or JanitorAI, you will likely see prompts like “You are talking to an AI” appearing on these platforms over the next few months, or new age verification processes. These are the practical implementations of this 14-state regulatory wave.
