A California bill that would regulate AI companion chatbots is close to becoming law

The California State Legislative Assembly took a big step toward AI control on Wednesday night, passed SB 243 – A bill that controls the AI ​​companion chatbots to protect minor and weak users. The law has passed with bilateral support and now the state has gone to the Senate for the final vote on Friday.

If Governor Gavin News signs the law on this bill, it will be effective in January 1, 2026, AI chatboat operators in California will be created as the first state to implement the protection protocol for AI companions and their chatboats will be legal to meet these values.

The bill is especially to prevent companion chatbots, which defines the law as an AI system that adapts, provides human-like reactions and is involved in conversation related to self-loss or sexually clear content capable of meeting user’s social needs. The bill will need platforms to provide repeat alerts in the bill – every three hours for minors – reminding them that they are talking to an AI chatboat, not a true person and they should take a break. It establishes annual reports and transparency requirements for AI companies, which provide chatboat with the main player OpenAE, character.AI and replica.

California bills will also be allowed to be permitted who believe that they were injured in the case of lawyer relief, damage (up to $ 1,000 per violation) and attorney fees against AI companies.

State Senator Steve Padilla and Josh Baker will visit the state Senate for the final vote on Friday, launched in January. If approved, it will go to the Governor Gavin News to sign the law, the new rules will be effective with the requirements of the required report, starting 1 January, 2026 and July 1, 2027.

The bill has achieved speed by following the California Legislative Assembly Teenage Adam Rhine’s deathWho committed suicide after a prolonged chat with Openai’s Chatzipt, which is involved in discussing and planning his death and self-loss. The law responded to the leakage Internal document It has been shown that Meta’s chatties were allowed to be involved in “romantic” and “sensual” chats with kids.

In recent weeks, US lawmakers and regulators have responded to intense investigation into the AI ​​platform to protect minors. The Federal Trade Commission AI Chatbots are preparing to investigate how children will affect children’s mental health. The Attorney General of Texas contains Ken Packstone Start investigation Meta and character. In the IA, they complain to misleading children demanding their mental health. Meanwhile, both Sen Josh Haoli (R-Mo) and Sen Ed Marki (D-MA) Meta has launched a separate probe.

Padilla told TechCrunch, “I think the loss is possible, which means we need to move quickly.” “In particular, minors know that they are not talking to a true man, when these platforms are thinking of hurting themselves or saying that they are in crisis, these platforms can be arranged to ensure that these platforms connect people with proper resources, we can arrange reasonable protection. [and] To ensure that inappropriate elements do not have inappropriate exposure. “

Padilla AI agencies also emphasized the importance of sharing their data by referring to users in crisis services every year, “So we have a better idea about the frequency of this problem than just being aware of the loss or bad.”

There was a more powerful requirement before SB23, but many were shaken by amendments. For example, the bill was essentially needed to prevent the AI ​​chatbots from using “variable rewards” strategies or additional features that encourage extra busyness. These techniques, used by AI associate agencies such as replica and character, provide special messages, memories, story lines or the ability to unlock the rare personality or new personality, which critics are called addictive rewards loop.

The current bill also removes the provisions that operates the trackbots of the operators to track and report how many times the trackbots or users begin to discuss with users.

Baker told TechCrunch, “I think it hurts the right balance to reach the victims without applying anything, which is impossible for companies, because it is not possible or not just a lot of papers for anything,” said Baker TechCranch.

SB243 is moving towards becoming SB243 Act at the time Silicon Valley companies are ing A few million dollars in pro-i Political Action Committees (PAC) to support the upcoming mid-melodic elections who are in favor of a light-touching approach for AI control.

The bill also comes when the weight of another AI protection bill in California, SB 53Which orders the need for a wide transparent report. Opeeni writes an open letter to the Governor News, asked him to leave the bill for the less strict federal and international structure. Main technology companies like Meta, Google and Amazon have also opposed SB 53. Only contrast The anthropologist says it supports SB 53The

Padilla said, “I reject this foundation that this is a zero amount of situation, that innovation and control are mutually exclusive,” Padilla said. “Don’t tell me that we can’t walk and chew gum we We can support innovation and development that we think is healthy and has the advantages – and this technology has the advantages of clearly – and at the same time we can provide reasonable security for the weakest people.”

TechCrunch Open AAE, ethnographic, meta, character AI and Replica reached for comments.