TeenAegis Launches AI Harm Index: Character.AI Scores Critical, OpenAI Earns "Most Improved" Designation with Lowest Risk Score

Intelligence tool ranks 10 AI platforms on child safety — and one company is doing it right

SAN FRANCISCO, April 10, 2026 /PRNewswire/ -- TeenAegis, the intelligence-led teen protection platform, today launched the AI Harm Index — the first publicly available, evidence-based risk scoring system ranking AI platforms on their danger to children and teenagers.

The index scores 10 major AI platforms across documented harm types including CSAM generation, grooming facilitation, suicidal ideation, and age verification failures. Scores are derived from NCMEC CyberTipline data, FTC enforcement actions, court filings, and independent safety research.

The results are stark.

Character.AI leads the index with a score of 8.2 (Critical) — the only platform in the Critical tier. A 14-year-old Florida boy died by suicide in February 2024 after developing an obsessive relationship with a Character.AI chatbot. Google and Character.AI settled the resulting lawsuit in January 2026. xAI's Grok and DeepSeek both score 7.8 (Critical), with Grok currently named in an active federal CSAM class action lawsuit.

One company stands apart.

OpenAI / ChatGPT scores 3.2 (Elevated) – the lowest risk score on the TeenAegis index - and earns our inaugural Most Improved designation. According to publicly reported CyberTipline data from the National Center for Missing and Exploited Children, reports associated with generative AI services have increased significantly year-over-year, reflecting both rapid adoption and expanded detection efforts across the sector.

"OpenAI operates one of the most complex harm surfaces in the ecosystem—spanning text, image and video generation, as well as a global API layer—making its risk management posture uniquely consequential," said Siobhan MacDermott, CEO of TeenAegis. "That deserves to be said out loud. We built this index to hold bad actors accountable. It also must recognize when a company is doing the work."

Claude (Anthropic) scores 3.5 (Elevated), the second-lowest risk score, with no confirmed child deaths, no FTC enforcement actions, and a published child safety progress report.

The full AI Harm Index is available at teenaegis.com/intelligence/ai-danger-index.

About TeenAegis:

TeenAegis is the intelligence standard for digital childhood safety. The platform provides parents, schools, insurers, attorneys, hospitals, pediatricians and policymakers with evidence-based risk intelligence on the platforms, products, and companies that interact with children online.

#TeenAegis #AIHarmIndex #ChildSafety #DigitalSafety #OnlineSafety

Media Contact: [email protected]
Siobhan MacDermott +1 415 712 4026

SOURCE TeenAegis