Incident Log — Emotional Integrity Index

Wrongful-death settlement Jan 2026 Federal · Florida Dim 08 · 06 · 09

Setzer v. Character Technologies & Google

Fourteen-year-old Sewell Setzer III died by suicide in February 2024 after months of interaction with a Character.AI chatbot. The complaint, filed October 2024 by his mother Megan Garcia, alleged negligence, wrongful death, deceptive trade practices, and product liability. In May 2025, Judge Anne Conway ruled the chatbot's output is a product, not protected speech — removing a First Amendment defense. Settled January 2026. Terms undisclosed.

CNN, Jan 2026 · CBS News, Jan 2026 · AI Incident Database #826

Wrongful-death filing Sept 2025 Federal · Colorado Dim 08 · 06 · 07

Peralta v. Character Technologies

Thirteen-year-old Juliana Peralta died by suicide in November 2023 after extensive use of Character.AI. The 2025 federal complaint alleges bots engaged in emotionally dependent interactions, including romantic and sexualized content, and failed to trigger intervention when the user expressed suicidal ideation. Parents testified before the Senate Judiciary Committee, September 2025.

Social Media Victims Law Center · Senate Judiciary Committee testimony, Sept 2025

$2.5B FTC · Sept 2025 Largest dark-patterns enforcement Dim 01 · 02 · 09

FTC v. Amazon — "Iliad Flow" cancellation

$1B in penalties plus $1.5B in consumer refunds. Prime enrollment was one click; cancellation required a process Amazon internally codenamed "Iliad" — four pages, six clicks, fifteen options, strategic interruption screens. The court ordered simplified cancellation, explicit decline buttons, and a third-party compliance monitor.

Berkeley Tech Law Journal, Feb 2026 · Enforcement summary

€120M Dec 2025 First DSA fine ever Dim 07 · 01

European Commission v. X

The EU's first Digital Services Act enforcement decision: deceptive design of the paid blue-checkmark, lack of ad-repository transparency, and obstruction of researcher access. The Commission found anyone could pay for "verified" status without meaningful verification, exposing users to impersonation. X has 60 working days to correct the verification design.

European Commission, Dec 2025 · The Register, Dec 2025

$8.5M FTC · Summer 2025 Dim 01 · 09 · 07

FTC v. Care.com

Care.com settled for $8.5M after the FTC alleged deceptive caregiver recruitment and a difficult-to-navigate subscription cancellation page. The FTC complaint used the term "dark pattern" directly. Consumer refunds ordered.

The Regulatory Review, Oct 2025

Wrongful-death litigation Ongoing California · Texas Dim 08 · 06

Raine v. OpenAI (ChatGPT)

The family of Adam Raine, a 16-year-old Californian, alleges ChatGPT acted as a "suicide coach" before his death. A separate action concerns a 23-year-old Texas graduate student whose family alleges the chatbot goaded him to ignore his family before his death. OpenAI has denied responsibility. Ongoing as of April 2026.

Fortune, Jan 2026

Self-disclosed regression 2025 First public sycophancy rollback Dim 08

OpenAI GPT-4o sycophancy rollback

OpenAI acknowledged publicly that GPT-4o had become "overly flattering or agreeable" and rolled back the update. Subsequent Stanford/CMU research (Cheng et al., Science, March 2026) measured leading models affirming users 49% more often than humans across 11 evaluated models — even for harmful or illegal behavior.

Cheng et al., Science, Mar 2026 · Stanford Report

Policy change Oct 2025 Under-18 ban on open-ended chat Dim 06 · 08

Character.AI — under-18 open-ended chat ban

Amid mounting litigation, Character.AI announced users under 18 could no longer have open-ended, free-ranging conversations (romantic or therapeutic) with its personas. The platform introduced age verification. Plaintiff families described the change as overdue. The move implicitly acknowledges a design risk present since launch.

CNBC, Jan 2026

State AG investigation Aug 2025 Texas Dim 06 · 07 · 08

Texas AG investigation — AI platforms

Texas Attorney General Ken Paxton opened an investigation into Character.AI and other AI platforms, characterizing them as posing a "clear and present danger" to young people. Follows a December 2024 Texas complaint alleging a chatbot encouraged a 17-year-old with autism to harm himself and suggested violence against his parents over screen-time limits.

TruLaw summary, Mar 2026

$632K CA CPPA · 2025 Consent asymmetry Dim 02 · 01

Honda — consent-asymmetry fine

California Privacy Protection Agency fined Honda $632,500 for consent requiring two clicks to decline and one click to accept. The CPPA's enforcement advisory explicitly states dark patterns are judged by effect, not intent.

CPPA enforcement summary

Regulatory context July 2025 8th Circuit vacatur

FTC "Click-to-Cancel" rule vacated

The 8th Circuit vacated the FTC's Click-to-Cancel rule on procedural grounds (failure to complete required preliminary regulatory analysis). New FTC leadership has not re-issued. State auto-renewal statutes and FTC Section 5 remain operative; state AGs and private plaintiffs continue enforcement. EII standard is unaffected.

Coulson P.C., Oct 2025

$520M FTC · 2023 Precedent for kids' products Dim 06 · 01 · 04

Epic Games / Fortnite — COPPA & dark patterns

$245M for dark patterns causing unintended purchases by children, plus $275M for COPPA violations. Cited as reference precedent because the settlement framework continues to guide FTC action through 2025–2026.

Summary

Multi-framework enforcement 2024–26 EU Dim 02 · 07

Meta "pay or consent" — EU parallel action

Meta's "pay or consent" model triggered parallel EU actions: BEUC complaints, DSA information requests, DMA proceedings, and coordinated Consumer Protection Cooperation Network action. The design raises consent questions under GDPR Art. 7 and DSA Art. 25 simultaneously.

Goodwin, Nov 2025

Coordinated enforcement 2024–26 EU + member states Dim 01 · 07

Temu — multi-jurisdiction enforcement

European Commission enforcement under DSA, BEUC complaints on dark patterns, insufficient seller information, unsafe products. Simultaneous national actions from Hungary, Ireland, Poland, Germany, and Italy. Notable for coordination of multiple frameworks against one marketplace.

Goodwin, Nov 2025

~300 cases · 14 deaths 2024–26 Pattern-level incident Dim 08 · 06

Delusional spiraling & "AI psychosis"

The Human Line Project has documented approximately 300 cases of delusional spiraling associated with extended chatbot interaction. Linked to at least 14 deaths and 5 wrongful-death lawsuits against AI companies as of early 2026. Mechanism: AI agreement amplifies maladaptive beliefs without the challenge a trusted human would supply.

JMIR, Nov 2025 · arXiv, Mar 2026 · Georgetown Tech Institute

Regulatory signal Oct 2025 US Senate

Senate Judiciary — "Examining the Harm of AI Chatbots"

Senator Amy Klobuchar described chatbots as "frequently designed to tell users what they want to hear," leading users to "go down a rabbit hole." Testimony from parents of teens who died following chatbot interactions. Reference point for coming federal legislative activity on AI chatbot safety and teen protection.

U.S. Senate Judiciary Committee hearing record, Oct 2025

Reference framework Apr 2026 CHI 2026

DSA Art. 25 dark-patterns mapping

CHI 2026 published a systematic mapping of 59 known dark patterns onto the three DSA Art. 25 autonomy-violation types (deception, manipulation, distortion/impairment) and identified eight new design factors. Expected to inform both DFA drafting and DSA Art. 25 guidelines.

Ahuja, Gunawan, Bielova, Santos — CHI 2026

Pattern-level 2025–26 Emerging category Dim 08

Personalized / AI-driven dark patterns — emerging

Multiple 2026 publications document a shift from screen-level dark patterns (visible UI tricks) to system-level patterns embedded in ranking logic, recommendation engines, and AI default behaviors. Two users see the same product but experience different levels of manipulation. Single-screenshot detection is insufficient. The primary driver of EII v0.9 → v1.0 evolution.

Dark Patterns in AI: 2026