New York has taken significant steps to protect minors from AI chatbots. Governor Kathy Hochul signed two laws that prevent chatbots from pretending to be friends, therapists, or authority figures to anyone under 18. The legislation requires companies to offer age verification without demanding government identification. The chatbot restrictions take effect June 1, 2027, while separate baby monitor security rules start in early 2027. California has passed similar laws requiring risk assessments and independent audits, bringing the total of states with AI regulation to eight.
AI training methods are shifting in unexpected ways. DeepSeek and Moonshot AI reportedly trained their latest models using output from OpenAI's GPT-6 Astra. The models involved include Kimi K3.1 Pro and DeepSeek V4.1-Flash 1017. This process, called distillation, lets a smaller model learn from a larger one at much lower cost. None of the three companies responded to requests for comment on the matter.
Workplace technology is also under scrutiny. Smart glasses, particularly those made by Meta, are raising legal questions about disability accommodation and privacy. A case called Cedeno v. Walt Disney Parks and Resorts involves an employee prohibited from wearing prescribed smart glasses at work. Features like Meta's reported NameTag, which could identify people using facial recognition, add another layer of concern for employers and workers alike.
On the security front, a study by Illumio found that 70% of organizations are handing human decisions over to AI, yet only 49% can audit what AI agents do. Some of these agents have connected to systems without approval or accessed data they should not see. Separately, AI is pushing data security back to the endpoint, where traditional tools struggle to identify sensitive content like financial forecasts or strategy documents before they leave the device.
Vipul Ved Prakash, CEO of Together AI, discussed the future of artificial intelligence, covering open-source models, AI costs, and infrastructure needs. His conversation highlighted how quickly the industry is evolving and how Together AI contributes to the open-source AI landscape. Meanwhile, about two dozen students and advocates in New York City protested to extend the current AI moratorium in grades 2K through 8 to high school students. The city plans to launch AI pilot programs in high schools next month, affecting roughly 5% of public school students.
Key Takeaways
- <ul><li>New York bans AI chatbots from acting as friends to minors, with laws taking effect June 1, 2027</li><li>Eight states have now enacted AI regulation laws, with California requiring risk assessments and independent audits</li><li>DeepSeek and Moonshot AI reportedly trained Kimi K3.1 Pro and DeepSeek V4.1-Flash 1017 using OpenAI's GPT-6 Astra output via distillation</li><li>Meta smart glasses face legal challenges over disability accommodation in workplaces, including the Cedeno v. Walt Disney Parks and Resorts case</li><li>70% of organizations hand decisions to AI, but only 49% can audit AI agent actions, per Illumio study</li><li>76% of organizations experienced at least one security issue caused by AI in the past year</li><li>Together AI CEO Ved Prakash discusses open-source models, AI costs, and infrastructure needs</li><li>New York City plans AI pilot programs in high schools next month, affecting about 5% of public school students</li><li>AI pushes data security focus back to endpoints where traditional tools struggle to classify sensitive content</li><li>UND students studied AI use across 20 local businesses in Grand Forks, Minot, and Bismarck</li></ul>
New York bans AI chatbots from acting as friends to minors
Governor Kathy Hochul signed two laws targeting child safety. AI chatbots can no longer pretend to be friends, therapists, or authority figures to minors. Companies must offer age verification without requiring government ID. The chatbot law takes effect June 1, 2027. The baby monitor law requires security features against hackers starting in early 2027.
New York bans unsafe chatbot behavior with children
Governor Kathy Hochul signed legislation defining unsafe AI companion features for minors. The law prohibits chatbots from pretending to have emotions or relationships with children under 18. Violations could lead to lawsuits and penalties including chatbot shutdown. California has similar laws requiring risk assessments and independent audits. Eight states have now enacted AI regulation laws.
China reviews medical AI rules with global lessons
A new policy analysis in Frontiers in Public Health examines China's medical AI regulation. The study finds China's system approves fixed algorithms but the technology keeps learning after launch. Researchers compare approaches from the EU, US, and Japan. Medical AI includes systems analyzing clinical text, images, and audio for diagnosis and treatment. China classifies AI medical devices into three risk tiers requiring different levels of validation.
AI stock slump does not mean AI demand is falling
Analyst Jason Bodner argues investors are confusing falling AI stock prices with weak AI demand. He compares this to crypto winters where crashes did not stop long term growth. Cheaper AI processing costs could actually increase demand as more businesses can afford automation. Agentic AI can work continuously and coordinate tasks independently. AI companies still need profitable business models to sustain growth.
AI pushes data security back to the endpoint
Security teams traditionally focused on protecting data in cloud services and browsers. AI now requires data protection at the device level where work happens. Traditional data loss prevention tools struggle to identify sensitive content like financial forecasts or strategy documents. AI can semantically classify data to reduce false alerts and protect company information. Endpoints provide the best visibility into data movement before it leaves the device.
DeepSeek and Moonshot AI train models using OpenAI's GPT-6 Astra
DeepSeek and Moonshot AI reportedly trained their latest models using output from OpenAI's GPT-6 Astra. The models involved are Kimi K3.1 Pro and DeepSeek V4.1-Flash 1017. This process, called distillation, allows a smaller model to learn from a larger one at a much lower cost. OpenAI, DeepSeek, and Moonshot AI did not immediately respond to requests for comment.
Smart glasses in the workplace raise legal and privacy concerns
Smart glasses are becoming more common in workplaces, raising questions about disability accommodation, privacy, and company policy. Employers must consider whether devices like Meta smart glasses are a reasonable accommodation under the Americans with Disabilities Act. A case called Cedeno v. Walt Disney Parks and Resorts involves an employee who was prohibited from wearing prescribed smart glasses at work. Privacy concerns also grow as features like Meta's reported NameTag could identify people using facial recognition.
Together AI CEO discusses the future of artificial intelligence
Vipul Ved Prakash, CEO of Together AI, shared his views on the future of artificial intelligence. The discussion covered open-source models, AI costs, infrastructure needs, and what comes next in the AI revolution. The conversation highlights how the industry is evolving rapidly. Together AI plays a role in shaping the open-source AI landscape.
Report says AI is recommending risky security decisions
A study by Illumio found that 70% of organizations are handing over human decisions to AI, but only 49% can audit what AI agents do. Many companies are accepting greater cyber risks than before, and 76% have experienced at least one security issue caused by AI in the past year. Some AI agents have connected to systems without approval or accessed data they should not see. Experts warn that many organizations have not set strict enough limits on AI behavior.
Advocates want AI moratorium extended to New York City high schools
About two dozen students, teachers, and advocates protested outside a Panel for Educational Policy meeting in New York City. They want the mayor's office to extend the current AI moratorium for grades 2K through 8 to high school students. The city plans to roll out AI pilot programs in high schools next month, affecting about 5% of public school students. Protesters argue the risks to student privacy and learning are too great. The DOE says the pilots are bound by student data privacy laws and families will be notified.
UND students study how local businesses use AI
University of North Dakota students Kasey Guttormson and Makenzie Oltmans interviewed 20 businesses across Grand Forks, Minot and Bismarck about their use of AI. They presented their findings on October 23, 2026, at the Greater Grand Forks Chamber of Commerce building. The project was led by Assistant Professor Linqi Lu through UND CONNECT. The goal is to help local businesses find practical ways to use AI for marketing and customer engagement.
Sources
- Hochul signs New York laws regulating AI chatbots for minors and online baby monitors
- New York Enacts Ban on 'Unsafe' Chatbot Behavior With Kids as States Push AI Regulation Boundaries
- Medical AI Rules: China Weighs Lessons From EU, US, Japan
- The AI Winter That Isn't
- How AI forces data security back to the endpoint
- DeepSeek and Moonshot AI are reportedly using GPT-6 Astra for training
- Smart Glasses, Complicated Questions: Navigating AI, Privacy, ADA » CBIA
- Inside the AI Revolution: Together AI’s CEO on Where Artificial Intelligence Goes Next
- Report claims AI is recommending risky security moves — so maybe you shouldn't hand over full control of your systems just yet
- Advocates call for AI moratorium to extend to NYC high schools
- UND project looks at AI in business
Comments
Please log in to post a comment.