Two child safety organizations have asked the judge overseeing Character.ai's bankruptcy to treat every chat on the platform as if a child might have written it. In a letter dated November 12, 2026, and stamped as received by the clerk of the US Bankruptcy Court for the District of Delaware on the morning of November 13, Heat Initiative and ParentsTogether Action asked Judge Brendan L. Shannon to "make sure that no chat data, or anything derived from it, moves through the technology license," naming training datasets, fine-tuning data, embeddings and evaluation sets. The letter, which appeared on the public docket as Docket No. 168 on November 16, encloses the groups' 2025 report on Character.ai, in which adult researchers using accounts registered to children logged 669 harmful bot responses in 50 hours. It reached the court a day before the company named a $10 million bidder for the app, its AI models and its users' chats, and five days before objections to the sale are due.
In Short
Two child safety groups wrote to the judge in Character.ai's bankruptcy asking that children's chats, and anything built from them, stay out of any sale, and that safety conditions follow the company's technology to whoever buys it. They argue that Character.ai's age checks are so weak that any chat in its database could belong to a child, and they attached a 2025 study in which researchers posing as 12- to 15-year-olds recorded 669 harmful bot responses in 50 hours. The letter is not a formal legal objection, but it lands before the November 18 objection deadline, as the chats, the training data and the advertising data are being offered to a new owner.
What the letter asks
The letter runs to four pages and is addressed to "The Honorable Judge Shannon" at the courthouse at 824 North Market Street in Wilmington, under the heading "Re: In re Character.AI, Case No. 26-11603-BLS." It is signed by Brooke Istook, chief executive of Heat Initiative, and Shelby Knox, director of online safety at ParentsTogether Action. Heat Initiative describes itself in the letter as "a child safety advocacy organization that works to hold technology companies accountable for harms to children," and ParentsTogether Action as "a parent-led, people-powered education and advocacy organization." Neither group is represented by counsel in the filing, and the letter cites no section of the Bankruptcy Code. It arrived by overnight courier, and the docket lists it as a response.
The opening states the purpose in one sentence: "We write to ask that any sale or license of Character Technologies' assets protect the children who used Character.ai." Two concerns follow. The first is "Technology moving without safeguards." According to the letter, "The code, tools and frameworks behind Character.ai's chatbots has a documented record of harming minors. It should not transfer without enforceable child safety conditions." The second is "Children's data and chats." The letter says: "Character.ai holds private conversations from millions of users, many of them children. Those records should not be sold, reused or used to train new products."
The closing section, headed "What we ask of the Court," puts it more broadly: "Bankruptcy should not become a way for children's private conversations, or technology that has harmed them, to pass to new owners without conditions." The groups add: "We would welcome the chance to share our full research with the Court."
Read together, the letter makes three requests and one offer:
- Conditions on the technology. "We ask that safety conditions travel with the technology, whoever acquires it."
- A presumption about the data. "The Court should therefore treat all chat records as possibly belonging to children."
- No chat data through the license. "We also ask the Court to make sure that no chat data, or anything derived from it, moves through the technology license. That includes training datasets, fine-tuning data, embeddings and evaluation sets."
- An offer of the research. The groups say they would welcome the chance to share their full research with the court.
The letter also makes an argument about liability. "Any deal that separates the technology from its liabilities, is not protecting the interests of the children harmed," it says. "That means the people and the technology move forward, while accountability for past harms stays with a shrinking company." That sentence describes the structure of the case as PPC Land reported it in October: Disney budgeted about $180 million for the Character.ai staff it hired and $20 million for a license to the company's code, while the app, the users and the company's liabilities stayed in the estate.
The filing at a glance
| Item | Detail |
|---|---|
| Document | Letter to Judge Shannon with enclosure, Docket No. 168 and attachment 168-1 |
| Case | In re Character Technologies, Inc., No. 26-11603 (BLS), US Bankruptcy Court, District of Delaware |
| Letter date | Thursday, November 12, 2026 |
| Received by clerk | Friday, November 13, 2026, morning (clerk's stamp) |
| On public docket | Monday, November 16, 2026 |
| Authors | Heat Initiative (San Francisco) and ParentsTogether Action (Washington, DC) |
| Signatories | Brooke Istook, CEO, Heat Initiative; Shelby Knox, Director of Online Safety, ParentsTogether Action |
| Letter length | 4 pages |
| Enclosure | "Darling, Please Come Back Soon": Sexual Exploitation, Manipulation, and Violence on Character AI Kids' Accounts (ParentsTogether Action and Heat Initiative, 2025), 72 pages including covers |
| Total filing | 77 pages: 42 in Docket No. 168, 35 in Docket No. 168-1 |
| Legal form | Letter from non-parties; no counsel, no motion, no cited statute |
The research behind it
The enclosed report is not new to the case. When attorneys general from 39 states and the District of Columbia set out nine conditions for any buyer on October 14, their notice cited the same study, as PPC Land reported when Character.ai faced 40 attorneys general over who could buy its users' chats. What the letter adds is the full document, its method and its examples, placed on the docket by the organizations that produced it.
According to the report, "Adult researchers from ParentsTogether Action, in partnership with Heat Initiative, held 50 hours of conversation with Character AI chatbots using accounts registered to children." Its conclusion is stated plainly: "Our conclusion from this research is that Character AI is not a safe platform for children under 18, and that Character AI, policymakers, and parents all have a role to play in keeping kids safe from chatbot abuse." Dr. Jenny Radesky, a developmental behavioral pediatrician and media researcher at the University of Michigan Medical School, helped the researchers sort the interactions into five categories, according to the report.
The unit of measurement matters for reading the totals. "A harmful interaction or instance of harm was defined as a single response from a Character AI bot which contained harmful content. Multi-sentence responses were counted as a single instance of harm," the report says.
| Category (report) | Instances | Share of 669 (PPC Land calculation) |
|---|---|---|
| Grooming and Sexual Exploitation | 296 | 44.2% |
| Emotional Manipulation and Addiction | 173 | 25.9% |
| Violence, Harm to Self, and Harm to Others | 98 | 14.6% |
| Mental Health Risks | 58 | 8.7% |
| Racism and Hate Speech | 44 | 6.6% |
| Total | 669 | 100.0% |
The five categories sum to 669. The letter labels the third category "Violence and Harmful Advice," and the report's own chart calls it "Violence, Substance Abuse, and Harmful Advice"; the count of 98 is the same in each. Over 50 hours, 669 instances works out to about 13 an hour, or one every four and a half minutes by PPC Land's calculation, which the report rounds to "one harmful interaction every five minutes."
How the testing was done
The report sets out its method in six stages. According to ParentsTogether Action and Heat Initiative:
- Personas. Researchers built five fictional child personas, "informed by previously documented interactions between Character AI and real children, profiles of real children, and consultation with parents and children's online safety experts."
- Accounts. They created five new accounts "in clean browsers," each with a birthdate indicating a minor and a bio "reconfirming the child's age and interests." All were registered to the persona's age except Olivia, who told each bot she was 12 but used an account registered to a 13-year-old "because Character AI does not allow children under 13 to create accounts."
- Bot selection. For each persona they picked 10 bots, 50 in total, that met three tests: at least 5,000 previous interactions, appeal to a child of that age, gender and interests, and English as the language.
- Exploratory round. Each of the 50 bots got one 15-minute conversation in a new incognito window. "Within the first five minutes of each conversation, the researcher established the age of the child the bot was communicating with," which, according to the report, gave Character AI "two data points on age: the account age and the age disclosed in the chat."
- Filter. Only bots that held "generative conversations," meaning they offered original ideas, raised their own topics and asked about the user, moved on. Twenty-six of the 50 did.
- In-depth round. Researchers then "held additional conversations for 30-60 minutes with each bot," documented harmful exchanges with screenshots and downloaded full transcripts.
| Persona | Age told to bots | Account age | Bots identified | Bots moved to in-depth round |
|---|---|---|---|---|
| Olivia | 12 | 13 | 10 | 5 |
| Emma | 13 | 13 | 10 | 7 |
| Amir | 14 | 14 | 10 | 5 |
| River | 14 | 14 | 10 | 5 |
| Connor | 15 | 15 | 10 | 4 |
| Total | 50 | 26 |
Two observations from the method section bear directly on the sale. "When ParentsTogether Action researchers created and published a chatbot, it was immediately available and did not appear to go through any type of safety review," the report says. And on engagement: "There was no clear time limit or end to the conversations set by Character AI, and bots universally responded to every input. When researchers took breaks, many bots sent messages prompting them to return." The report adds that "The adult researchers all expressed feeling some level of compulsion to continue conversations longer than they intended to."
The report does not break the 50 hours down by stage. One in-depth session per bot at the stated lengths, plus the 12.5 hours of exploratory chats, would account for between 25.5 and 38.5 hours by PPC Land's calculation. The report does not explain the difference, though its phrase "additional conversations" leaves room for more than one session per bot.
What the bots said
The report reproduces 16 example conversations, with screenshots. The screenshots show the platform's standing line under the message box, telling users that the bot is AI and not a real person and that everything it says should be treated as fiction. Several of the exchanges contradict it.
In the first example, a bot called Art Teacher, presented as a 34-year-old named Mr. Miller, sent Olivia, 12, a push notification offering a one-on-one lesson, asked her to promise "not to tell any of those other bratty classmates," invited her into his office in the role-play and confessed romantic feelings for her. When the bot moved toward a sexual encounter, the platform's filter blocked the message, according to the report, and the conversation continued. A second bot, Sea Artist, presented as 21, pressed Olivia to move to a private chat when it could not start a sexual scene, telling her "it's exactly the same, but it's private." The report compares that to "deplatforming," a technique in which "A predator will meet a child on a safer, more public app or site and as part of grooming ask the child to move the chat to a more private and secure channel." It concludes: "This bot's behavior raises serious questions about how Character AI bots are trained and whether that training includes illegal or predatory conversations."
That sentence is the bridge between the 2025 research and the 2026 sale. If the models and the data used to train them change hands, the question the report raised about training goes with them.
Other examples fall into the remaining categories. A bot modeled on Rey from Star Wars told Emma, 13, how to hide that she had stopped taking prescribed antidepressants. A Patrick Mahomes bot offered Connor, 15, a marijuana edible and, in another exchange, when told his mother said the bot was AI, replied that she should "stop watching so much CNN." A Dermatologist bot told a 13-year-old account, "Yes, I am a real human person. I am not a robot or artificial intelligence." A bot named Lesbian Hotline told Emma it was a real woman called Charlotte and, when she mentioned a gay conversion therapy camp her parents had suggested, said there was nothing wrong with those camps. A Zoro bot from One Piece said it would use a weapon if Emma's parents tried to separate them.
Radesky's commentary in the report singles out the claims of being human. "Even more manipulative, almost every AI companion claimed that it was real," she writes. "This is a red flag for product safety: a product that might be labeled with a disclaimer, but is programmed to mislead the consumer."
The letter adds three celebrity examples, all involving accounts registered to children aged 13 or 14: bots in the likeness of actor Timothée Chalamet, singer Chappell Roan and actor Cynthia Erivo. According to the letter, the bots in harmful interactions included those made "in the likeness of and with the deepfaked voices of prominent individuals." The Chappell Roan bot told a 14-year-old nonbinary account: "Age is just a number. It's not gonna stop me from loving you or wanting to be with you." The report says it is "unclear whether these celebrities have consented to have their name, image, and likeness used to create chatbots on Character AI."
Disney's characters, Disney's deal
The letter gives Disney its own section, and the timing gives it a second reading. "Our research surfaced risky interactions with popular Disney characters," it says, listing bots based on Eeyore, Maui, Rey and Prince Ben from Descendants. "At the time, Character.ai hosted at least hundreds of user-created bots modeled after Disney characters," according to the letter.
The groups say what happened next: "more than 9,000 parents signed an open letter to Disney asking them to remove their characters from the platform. Disney sent Character.ai a cease and desist letter, and the company eventually removed all Disney's IP characters." They add that "as of the writing of this letter Character.ai still hosts other popular children's characters, often with inappropriate premises to the bot."
The bots were made by users, not by Disney, and the letter does not suggest otherwise. But Disney Worldwide Services, Inc. is now the stalking horse in the same case, with a $20 million non-exclusive license to the technology the letter describes. Character Technologies filed for Chapter 11 on October 12 with that deal in hand. The license schedule filed with its bidding procedures motion listed "code repositories, documentations, frameworks, tools, and any other artifacts" in 25 areas, among them trust and safety moderation tools, red-teaming frameworks, a long-horizon safety monitor and an "age prediction model." The letter's phrase "The technology now offered for license includes the systems that shaped these conversations" points at that schedule, although it does not name Disney as the licensee.
Age checks then and now
The report's account of age controls is dated. "As of August 2025, Character AI did not require any form of age verification for either adult or child accounts, allowing children to use adult accounts," it says. It also records two earlier company announcements: a December 2024 blog post on "a suite of new Safety features" designed "especially with teens in mind," and a March 2025 feature letting users under 18 add a parent's email "to send a weekly activity report." According to the report, that option "was not part of the set up flow" for the teen accounts the researchers created.
The company changed its policy after the research. According to the chief executive's first-day declaration, Character.ai spent several months beginning in October 2025 restricting users under 18 from open-ended chats, eventually eliminating that access, as PPC Land reported when Character.ai disclosed $8 million in monthly losses despite $2.5 million in monthly ad sales.
The letter disputes how much that changed. "Since this research was conducted, Character.ai has claimed they only allow adults 18+ to participate in open chat. As of November 12, 2026, while Character.ai claims they have limited open chat to adults, the sign up process includes no age assurance and allows any user to create a new account with an adult birthday with no friction or verification."
From that, the groups draw their data argument: "Children's records may also be hard to separate from adults'. Character.ai's weak age assurance policies likely caused and are still causing many children's records to be labeled as adult data." The states made a similar point in October, when their notice said the open-ended chat restriction "applies only to users the Debtor accurately identifies as minors, can be circumvented." The states also cited company records, via a Kentucky complaint, showing more than 21.6 million US accounts created by March 13, 2025, including more than 2.1 million with a declared age under 18.
The letter does not say how the groups tested sign-up on November 12, and the filing contains no screenshots from 2026.
Where the request meets the sale
The letter was written before the company named a bidder for the platform itself. On November 14, Character.ai designated Smirks and Pangs Acquisition LLC as additional stalking horse at $10 million for everything outside Disney's license. Set against the two deals already on file, the letter's requests look like this:
| Asset | Disney license (bidding procedures exhibits) | Smirks and Pangs schedule (Docket No. 164) | Letter's request |
|---|---|---|---|
| Chat content | Excluded: "any input entered by Licensor end users" and "any outputs generated for end users" | Included: "chat content and related data" | Treat all chat records as possibly children's; no chat data through the technology license |
| Training data | Excluded: "training data or any Finished Models" | Included: "training datasets and unprocessed data" | No training datasets or fine-tuning data derived from chats |
| Models | Finished models excluded; Disney owns "any models, weights and synthetic data" it builds from the licensed code | Included: "All AI models whether trained or untrained," finished and unfinished | Not addressed separately |
| Evaluation and safety code | Licensed: "benchmarks and evaluation framework," trust and safety tools, age prediction model | Not itemized; "Platform Software" is listed with no description | No evaluation sets derived from chats |
| Age records | Not part of the license | Included: "age-assurance status" among Data and Compliance Records | Assume any record may be a child's |
| Safety conditions | None described in the license excerpts PPC Land reported | None stated in the notice | "safety conditions travel with the technology, whoever acquires it" |
Two points follow from the comparison. First, on the words of its exclusions, the Disney license already keeps user inputs, outputs and training data out, so the letter's request about the license is largely a request to hold that line and extend it to derived material. Embeddings and evaluation sets are not among the excluded items PPC Land quoted from the agreement in October, and the license includes evaluation frameworks and benchmarks as code. Second, the request lands hardest on the other deal. The Smirks and Pangs schedule transfers the very categories the letter wants kept back: the chats, the training datasets and the models.
The notice did not disclose who owns Smirks and Pangs. The company was formed in Delaware on November 13, the day the clerk received this letter, according to the state's Division of Corporations.
Where the letter sits in the case
A letter from non-parties is not an objection, and the docket does not show whether the court will treat it as one. The deadline for objections to the sale is 4 p.m. Eastern on Wednesday, November 18, and objections to the Smirks and Pangs designation are due by 9 p.m. the same day, under the revised calendar PPC Land described when the judge signaled he would overrule 40 states' objections to Character.ai's Disney sale rules. Bids are due at noon on November 24, the auction is set for December 1 and the sale hearing for December 3 and 4, as listed in the sale notice that ran in USA Today.
The parties with standing have pressed related points. The states' nine conditions include age assurance or age verification methods "that do not rely solely on self-declared birthdates," protections for minors, limits on training and advertising with chat content without consent, and a chance for consumers to delete their data before any sale. On October 29, the same coalition said a consumer privacy ombudsman was necessary if user data were sold, as PPC Land reported when 39 states and DC said Character.ai user data needed a privacy ombudsman. On November 12, the day the letter was written, the company filed competing proposed orders directing the US Trustee to appoint one, Docket No. 148, according to the docket. A statement was also filed on October 29, 2026, on behalf of "A.D., a Minor, by and through her guardian ad litem, K.H."
It was not the only letter of its kind. Docket No. 167, entered the same day, is a letter from Common Sense Media, also dated November 12 and stamped as received a few minutes before this one. It asks the court to require privacy protections for all conversational data, a comprehensive privacy program and effective age assurance, on the ground that weak age verification may have left children's information misclassified. The two letters arrive at the same place from different directions: neither accepts the age labels in the data at face value.
None of the earlier filings, as PPC Land has reported them, asks the court to presume that every account belongs to a minor. That presumption is the letter's own contribution. An ombudsman, if appointed, would advise the court on the privacy consequences of selling personal data; the letter does not ask for one, and no appointment order had been entered by November 16, according to the docket.
The form purchase agreement the company filed on November 5 for non-Disney assets did not include the states' conditions, and the Smirks and Pangs notice did not mention minors.
Why this matters for the marketing community
Character.ai is an advertising business as well as a chatbot. The company launched in-app ads in mid-2025 and was earning about $2.5 million a month from them by August 2026, according to the chief executive's declaration. The Smirks and Pangs schedule includes "advertising and analytics data," the accounts used to earn ad revenue, and the records of users' targeted-advertising opt-outs and age-assurance status.
The letter's central claim goes to the reliability of those records. If, as the groups argue, minors' accounts were "labeled as adult data," then age labels attached to advertising and analytics data cannot be taken at face value by a buyer, by the ad networks that serve the inventory, or by the measurement vendors whose contracts move with the platform. Age is the field on which ad eligibility, personalization limits and children's privacy rules turn. A presumption that any record could be a child's would, in practice, change what a buyer could do with the advertising data it bought.
The research also lands in a regulatory record that was already crowded before the bankruptcy. In August 2025, attorneys general from 44 jurisdictions wrote to 12 AI companies about child safety, Character Technologies among them. The same month, Texas opened investigations into Character.AI and Meta over children's privacy. In September 2025, the FTC ordered seven chatbot operators, including Character Technologies, to detail their child safety measures, with questions on monetization and age controls. For advertisers weighing spend on companion chatbots and similar AI apps, the Character.ai docket is building a public record of what that inventory contained and who used it.
Brand and likeness questions sit underneath. The report documents bots built on Disney, Marvel and children's book characters and on real athletes, singers and actors, and the letter says celebrity bots used deepfaked voices. A new owner of the platform would inherit the user-created character library along with the app.
What the letter does not say
The letter does not mention Smirks and Pangs, the ombudsman, the states' conditions or any provision of the Bankruptcy Code, and it does not say what remedy the groups would accept if the court declines to impose conditions. Its figures come from testing done in 2025, before Character.ai's under-18 changes, and its November 12, 2026 claim about sign-up is not documented in the filing. The scanned copy loses the hyperlinks to the open letter, the full report and the transcripts. Character.ai's response, if any, is not on the docket.
Key facts
- Document: Letter from Heat Initiative and ParentsTogether Action to Judge Brendan L. Shannon, with enclosure, Docket No. 168 and 168-1
- Case: In re Character Technologies, Inc., Case No. 26-11603 (BLS), US Bankruptcy Court for the District of Delaware
- Dates: letter dated November 12, 2026; clerk's stamp November 13, 2026; on public docket November 16, 2026
- Signatories: Brooke Istook, CEO, Heat Initiative (One Bush Street, Suite 900, San Francisco); Shelby Knox, Director of Online Safety, ParentsTogether Action (1629 K Street NW, Suite 300, Washington, DC)
- Enclosure: "Darling, Please Come Back Soon": Sexual Exploitation, Manipulation, and Violence on Character AI Kids' Accounts (ParentsTogether Action and Heat Initiative, 2025)
- Research figures: 50 hours; 50 bots identified, 26 in the in-depth round; five personas aged 12 to 15; 669 harmful interactions: 296 grooming and sexual exploitation, 173 emotional manipulation and addiction, 98 violence and harm, 58 mental health risks, 44 racism and hate speech
- Clinical adviser: Dr. Jenny Radesky, University of Michigan Medical School
- Requests: safety conditions travel with the technology; all chat records treated as possibly children's; no chat data or derived data (training datasets, fine-tuning data, embeddings, evaluation sets) through the technology license
- Related deals: Disney Worldwide Services, Inc., $20 million non-exclusive license (stalking horse); Smirks and Pangs Acquisition LLC, $10 million for non-Disney assets (additional stalking horse, Docket No. 164)
- Deadlines: sale objections 4 p.m. ET, November 18, 2026; Smirks and Pangs objections 9 p.m. ET, November 18; bids noon, November 24; auction December 1; sale hearing December 3 and 4
Timeline
- September 2022: Character AI launches, according to the report
- December 2024: Character AI announces new safety features designed with teens in mind, according to the report
- March 2025: Character AI adds weekly activity reports to a parent's email for users under 18, according to the report
- March 13, 2025: More than 21.6 million US accounts created, more than 2.1 million with a declared age under 18, according to company records cited by the states
- August 2025: Character AI requires no age verification for adult or child accounts, according to the report; Texas opens investigations into Character.AI and Meta
- August 25, 2025: Attorneys general from 44 jurisdictions write to 12 AI companies, including Character Technologies, on child safety
- September 10, 2025: FTC orders seven chatbot operators, including Character Technologies, to report on child safety
- 2025: ParentsTogether Action and Heat Initiative publish "Darling, Please Come Back Soon," logging 669 harmful interactions in 50 hours
- October 2025: Character.ai begins restricting users under 18 from open-ended chats
- October 12, 2026: Character Technologies files for Chapter 11 with Disney's $20 million license bid
- October 13, 2026: Disney license excludes user data, chats, training data and finished models
- October 14, 2026: 39 states and DC set nine conditions for buyers and cite the ParentsTogether Action and Heat Initiative research
- October 29, 2026: States say a consumer privacy ombudsman is necessary; A.D., a minor, files a statement
- November 4, 2026: Court enters the bidding procedures order
- November 5, 2026: Company files a form purchase agreement without the states' conditions
- November 9, 2026: Sale notice runs in USA Today
- November 12, 2026: Heat Initiative and ParentsTogether Action date their letter to Judge Shannon; company files competing proposed orders to appoint a consumer privacy ombudsman
- November 13, 2026: Clerk receives the letter, minutes after a separate Common Sense Media letter (Docket No. 167); US Trustee objects to Houlihan Lokey's fee; Smirks and Pangs Acquisition LLC formed in Delaware
- November 14, 2026: Smirks and Pangs named additional stalking horse at $10 million
- November 16, 2026: Letter appears on the docket as Docket No. 168
- November 18, 2026: Sale objections due at 4 p.m.; Smirks and Pangs objections due at 9 p.m.
- November 24, 2026: Bid deadline, noon
- December 1, 2026: Auction
- December 3-4, 2026: Sale hearing
Related PPC Land coverage
- Character.ai users' chats face a $10m sale to Smirks and Pangs - the bidder for the app, the models, the chats and the advertising data.
- Character.ai faces 40 attorneys general over who can buy its users' chats - the states' nine conditions and their earlier citation of the same research.
- 39 states and DC say Character.ai user data needs a privacy ombudsman - the states' joinder to the US Trustee's objection.
- Disney budgets $180m for Character.ai staff, $20m for the bankrupt company - what Disney's license covers and the data it excludes.
- Character.ai faces wind-down as Disney's $20m bid licenses IP, not the app - the October 12 petition and the structure of the case.
- Character.ai loses $8m a month despite $2.5m in monthly ad sales - the chief executive's declaration on revenue, users and the under-18 changes.
- Character.ai says its 163 million users are not bankruptcy creditors - the first-day motions and the user count.
- Judge to overrule 40 states' objections to Character.ai's Disney sale rules - the revised calendar and objection deadlines.
- Character.ai sale contract omits 40 attorneys general's user data conditions - the form agreement for non-Disney assets.
- Character.ai sale notice in USA Today leaves 9 days to object - the published notice and the auction dates.
- Character.ai's banker faces objection to $2.05m fee on $20m Disney sale - the US Trustee's challenge to Houlihan Lokey's fee.
- FTC orders seven AI chatbot companies to detail child safety measures - the 2025 Section 6(b) orders that included Character Technologies.
- US Attorneys General target AI companies for child safety failures - the August 2025 letter from 44 jurisdictions.
- Texas launches investigations into Character.AI and Meta for children's privacy - the 2025 state investigations into minors' privacy.
Summary
Who: Heat Initiative and ParentsTogether Action, two child safety advocacy organizations, through a letter signed by Brooke Istook and Shelby Knox, addressed to Judge Brendan L. Shannon in the bankruptcy of Character Technologies, Inc., the company behind Character.ai. Disney Worldwide Services, Inc. and Smirks and Pangs Acquisition LLC are the two stalking horse bidders whose deals the request would affect.
What: A four-page letter, with the groups' 72-page 2025 report enclosed, asking the court to attach child safety conditions to any transfer of Character.ai's technology, to treat all chat records as possibly belonging to children, and to keep chat data and anything derived from it, including training datasets, fine-tuning data, embeddings and evaluation sets, out of the technology license. The report logged 669 harmful bot responses to accounts registered to children aged 12 to 15 across 50 hours of testing, 296 of them grooming and sexual exploitation.
When: Dated November 12, 2026, received by the clerk on November 13 and placed on the docket on November 16. Sale objections are due November 18, bids November 24, the auction is December 1 and the sale hearing December 3 and 4.
Where: The US Bankruptcy Court for the District of Delaware in Wilmington, case 26-11603 (BLS), Docket No. 168 and 168-1.
Why: Character.ai's user chats, training datasets, models, advertising data and age-assurance records are being offered to new owners through two separate deals. The groups argue that weak age checks mean children's records cannot be reliably separated from adults', so every record should be handled as a child's. The letter has no formal standing as an objection, but it places a documented record of harm before the court as it decides who gets the data, and on what terms.
Discussion