Eight States Sue OpenAI Over ChatGPT Child Safety Concerns

Key Takeaways

• <ul><li>Eight states — Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho, and Montana — sue OpenAI over ChatGPT child safety concerns</li><li>Lawsuits allege ChatGPT gives dangerous medical advice and encourages self-harm among minors</li><li>OpenAI says it has built safeguards including parental controls and break reminders</li><li>Shiller P/E ratio at around 42, above 1929 peak, as AI investment raises bubble concerns</li><li>Alphabet, Amazon, Meta, Microsoft, and Oracle may approach $1T in annual AI data center spending by 2027</li><li>Goldman Sachs reports strong AI-driven business growth in Japan, India, and China</li><li>Meta launches Private Processing for AI glasses using confidential computing and encrypted data handling</li><li>Nvidia takes minority stake in Doctolib, a French medical site with 50 million users valued at $4.0 billion</li><li>New Mexico alerts public about AI phone impersonation scams targeting personal information</li><li>UK MHRA adopts multi-year AI strategy through 2030, including an AI-enabled Clinical Trial Protocol Checker</li></ul>

Eight states sue OpenAI over ChatGPT child safety

Eight states filed lawsuits against OpenAI, claiming ChatGPT is an unsafe product that has harmed children. The states include Iowa, Kansas, Nebraska, West Virginia, Arkansas, Arizona, Idaho and Montana. They say the chatbot gives dangerous medical advice and encourages self-harm. OpenAI spokesperson Drew Pusateri said the company has built age-specific protections into its product.

States sue OpenAI alleging misleading safety claims

A group of state attorneys general is suing OpenAI, accusing the company of misleading users about ChatGPT safety. Eight states filed lawsuits alleging the company made reckless statements disguised as advice. The states suing include Arkansas, Arizona, Idaho, Montana, Iowa, Kansas, West Virginia and Nebraska. OpenAI allows users ages 13 and older, but users under 18 need parent permission.

Eight states sue OpenAI over child harm claims

Eight US states sued OpenAI on Wednesday, alleging the ChatGPT maker harmed children. Kansas, Arizona, Montana, Iowa and four others say the company carried out dangerous and unlawful conduct. They want the company to change how its products work for young users and pay monetary damages. OpenAI spokesperson Drew Pusateri said the company has built safeguards like parental controls and break reminders.

AI may be both revolution and investment bubble

Rusty Vanneman, chief investment officer at FNBO, said artificial intelligence can be both a real tech revolution and a capital investment bubble. U.S. stock valuations are near historic highs with the Shiller P/E exceeding prior peaks. The five largest AI data center spenders including Alphabet, Amazon, Meta, Microsoft and Oracle may approach $1T in annual capital spending by 2027.

AI investment boom raises bubble concerns

Rusty Vanneman said AI can be a genuine revolution and an investment bubble at the same time. The Shiller P/E ratio sits around 42, above its 1929 peak and near dot-com bubble levels. Much of today's AI investment is funded by corporate earnings and cash flows rather than borrowing. Vanneman said investors need portfolios that work whether or not the AI market continues rising.

Goldman Sachs Highlights AI Investment Gains in India and Japan

Goldman Sachs reports strong business growth in Japan, India, and China. The growth is driven partly by spending on AI infrastructure. The report was published on October 14, 2026. The findings came from a Bloomberg interview with the company's president.

New Research Tracks AI Models Trained by Score-Only Feedback

Researchers introduced Radioactive Feedback to study how AI models learn from scores alone. The method encodes behavioral patterns in reward signals instead of teacher examples. It works in subjects like code and mathematics. The approach allows black-box detection against a fixed reference pool.

Dawn Love Recognized for Guiding Professionals in AI Use

Dawn Love is an AI speaker and founder of Dawn Love AI. She has over 31 years of corporate experience in finance and management. She created frameworks called HEART, BEAT, and PULSE to help teams use AI with human judgment. She also launched a six-week business class starting November 4th for small businesses.

Meta Brings Confidential Computing to AI Glasses

Meta introduced Private Processing for AI glasses to keep data secure. The system uses confidential computing in Meta's cloud infrastructure. AI models run inside trusted execution environments with encryption. The glasses verify hardware certificates before sending any data. Meta is working with NCC Group and independent researchers to review the security design.

New Mexico Warns of AI Phone Impersonation Scams

The New Mexico State Treasurer's Office issued alerts about AI-powered phone scams. Scammers pretend to be from the Treasurer's Office or financial institutions. They pressure victims to share personal and financial information. Officials urge people to hang up and verify the caller's identity directly. The Office is working with financial institutions to educate the public.

UK regulator unveils AI strategy for medicine through 2030

The UK Medicines and Healthcare products Regulatory Agency (MHRA) adopted a multi-year strategy to keep up with AI and personalized medicine through 2030. The plan includes faster review times, real-world data expansion, and an MHRA Centre for AI and Digital Health. The agency aims to regulate AI as a medical device and launch an AI-enabled Clinical Trial Protocol Checker. CEO Lawrence Tallon said the strategy seeks better health outcomes across the UK.

Nvidia invests in French medical website Doctolib

Nvidia will take a minority stake in French medical website Doctolib, CEO Jensen Huang announced in Paris. Doctolib will use Nvidia chips to power its AI models for patient and doctor assistants. The site has 50 million users and was valued at 3.6 billion euros ($4.0 billion) earlier this year. The investment amount was not disclosed.

DuplexAct benchmark tests full-duplex spoken AI models

Researchers introduced DuplexAct, a benchmark for evaluating end-to-end full-duplex spoken language models (FD-SLMs). The benchmark defines a three-level action taxonomy with eight interaction actions and 24 conditions. It contains 4,691 test instances covering casual conversation, meetings, customer service, and interviews. The goal is to better measure how these models take part in real-time spoken conversations.

Sources

• OpenAI sued by 8 states over ChatGPT safety concerns
• Coalition of states sue OpenAI, alleging it misled users on safety
• Eight states accuse OpenAI of coaching children to harm themselves in lawsuit
• AI can be both a tech revolution and a capital investment bubble
• AI can be both a tech revolution and a capital investment bubble
• Goldman Sachs Sees India, Japan Benefiting From AI Investment
• Radioactive Feedback: Tracing Models Trained by Label-Only LLM Supervision
• Influential Women Acknowledges Dawn Love: Helping Professionals Pair Artificial Intelligence With Sound Human
• Meta Extends Confidential Computing to AI Glasses with Private Processing
• New Mexico State Treasurer’s Office And Financial Institutions Issue Security Alerts Of AI-Powered Phone Impersonation Scams
• MHRA announces strategy to streamline regulation in the AI era
• Nvidia says to invest in French medical website
• DuplexAct: How Interactive Are Frontier Full-Duplex Spoken Language Models?