The Deep View
Posts
China bans chatbot love over more than AI safety

China bans chatbot love over more than AI safety

The Deep View
July 16, 2026

Welcome back. OpenAI’s first device may be a smart speaker, a safer bet than wearables and a less legally fraught path amid Apple’s lawsuit. In New York, a first-of-its-kind pause on hyperscale data centers shows how power costs, water use, noise control, and local opposition could become hard limits on AI’s growth. And China’s ban on romantic chatbot companions reflects more than AI safety concerns. It intersects with larger societal impact issues that AI will have to navigate as its influence expands. —Jason Hiner

1. China bans chatbot love over more than AI safety

2. NY data center ban flashes a warning sign for AI

3. OpenAI's safer hardware path avoids Apple friction

CULTURE

China’s chatbot love ban goes beyond AI safety

China just made it illegal to fall in love with chatbots.

On Wednesday, the world's most populous nation and second-largest economy enacted new legislation and regulations that:

Make virtual relationships with chatbots illegal for minors
Require AI companies to alert a minor's emergency contact if an emotional crisis is detected
Bans companies from building chatbots designed to be virtual companions

Chinese tech giants Alibaba and ByteDance informed users on Wednesday that some features would be disabled, due to the new legislation. And Bloomberg reports that there are people in China grieving the impending "death" of their virtual companions.

In the US, New York and California have both enacted legislation putting limits on companion chatbots. But the Chinese government went a step further with full bans.

This comes at a time when, for the first time, more people globally trust China to do good in the world than the US, according to new data from Pew Research.

Of course, China's regulation of companion chatbots is partly due to AI safety and partly due to population decline. In 2025, China's population shrank for the fourth year in a row as the country's birthrate hit an all-time low.

Beyond banning chatbot relationships, China is also offering child-care subsidies and it has ended its decades-long one-child policy that was aimed at population control.

In general, the country wants to encourage young people to be in real-life relationships and minimize the distractions and drawbacks that could keep them from starting families.

There are also forces within Chinese society that have deep mistrust of AI, similar to the US. For example, Chinese regulators warned in a September 2025 policy document that AI could transform "traditional views on employment, fertility, and education" that could undermine the country's core values.

It doesn't take intense reflection to see the challenges that could result from chatbot companions that imitate human behaviors. With the US federal government continuing to hold back on broader AI legislation, expect to see more US states leading the way on guidelines for this and other topics in AI regulation. In light of Demis Hassabis's call for broader AI safety guidelines and a public-private partnership, the topic of AI ethics and societal impact continue to gain momentum. What's needed next is broader dialogue so that more people in more places can be part of the conversation around the ways we want to control and channel the capabilities of the technology.

TOGETHER WITH CRUSOE

Crusoe just launched Serverless Fine-Tuning

A complete path from proprietary data to production-ready models on purpose-built AI infrastructure — without the overhead of managing it.

Select a base model from a curated library of top open models, upload your dataset in JSONL or Parquet, configure with pre-configured best practices, and submit via UI, SDK, or API.

Jobs run with automatic recovery and restart, checkpoints save at every step, and your fine-tuned weights are yours to deploy anywhere.

Start fine-tuning →

POLICY

NY data center ban flashes a warning sign for AI

As public scrutiny mounts over the impact of data centers on communities, New York has taken swift action.

New York State Gov. Kathy Hochul signed an executive order on Tuesday banning for up to a year the construction of "hyperscale" AI data centers, or those that use more than 50 megawatts of power. The moratorium is the first of its kind in the nation.

The executive order also states that data centers will be required to either produce their own energy or pay a premium to tap into grid power, and notes that data centers will never be exempt from local zoning and approvals.

The moratorium will allow New York to establish "the strongest possible framework" and guardrails to reduce strain on the energy grid and protect communities and natural resources, Hochul said. Additionally, the state is exploring the requirement that data center operators pay into a larger fund to support the grid.

Hochul said in a press release that the ban intends to address the increasing power costs, impacts to the water supply, and noise pollution that come with these facilities.

"The bottom line is progress shouldn't arrive with a higher utility bill, deleted water supply or noise pollution," Hochul said. "So we have no choice but to address these challenges created by these massive facilities."

The ban comes at a time when public opinion has largely turned against AI data centers. A Gallup poll from May found that an average of 7 in 10 Americans oppose the construction of data centers in their area, with nearly 50% polling that they strongly oppose the local construction of these sites. Many have raised concerns about the environmental impacts, power constraints and quality-of-life impacts.

Meanwhile, companies like Meta have been trying to spin data centers as wins for the community, like its recent 5GW AI factory in Richland Parish, Louisiana. The company pitched this as "Teachers and Local Businesses Win as Meta Expands Louisiana Data Center."

Will New York's ban of AI infrastructure become a harbinger for the US more broadly. Data center pushback is just one manifestation of the larger problem that AI has: Despite Silicon Valley's fervor for the tech, public opinion towards AI has turned negative, with one poll showing that 55% of Americans broadly believe that it will do more harm than good. Data center discontent is particularly high because these facilities offer a physical display of the broader problem, representing something that many believe is a threat to their community, driving up utility bills, increasing pollution and tainting their water supply. And though tech executives like to parrot the narrative that the AI transformation is inevitable, the negative public perception of the tech could threaten the industry's future progress. Without buy-in from the broader public and communities becoming comfortable coexisting with data centers, the industry, which is grappling with a massive compute crunch, is limited in how big it can grow and how well it can compete with China.

TOGETHER WITH CORESIGNAL

10 years of data expertise. 50% off your first month.

Coresignal turns 10 this year, and you get the gift.

For a decade, we've been providing fresh company, employee, and jobs data for 1000+ organizations. Now's the best time for you to put this data to work and build your AI search, sales tech, or HR tech products.

Use code BDAY50 for 50% off your first month's subscription at checkout on our self-service platform.

Limited-time offer
Only valid for monthly subscriptions
New customers only

Claim your 50% discount

HARDWARE

OpenAI’s first device may start where Alexa stalled

OpenAI’s highly anticipated hardware project might not be as flashy as you expected.

A Bloomberg report, citing people familiar with the matter and published on Wednesday, detailed what the first OpenAI hardware product will look like: "a mobile, screen-free smart speaker designed to be a new type of home computer for the AI era."

Similar to an Amazon Echo, the speaker would function as an AI companion that lives in the home and helps control smart home devices, plays music, answers questions, and performs like the voice assistants from the era of Alexa, Google Home, and Apple HomePod.

The distinguishing features of OpenAI's AI speaker compared with past smart home speakers lie in the AI models powering it: ChatGPT and GPT-Live. Smarter AI could make it increasingly personalized over time, anticipating user needs and proactively surfacing information. It will also have a personality to make it feel more "human-like" and less mechanical, which OpenAI considers its most defining feature, according to the report.

Despite being a speaker, it is meant to be portable with a rechargeable battery so it can be moved around the house. It will also have a camera and other sensors to help it understand a user’s surroundings. Jonny Ive’s studio LoveFrom, is still helping design the speaker. This is only one of the five products that OpenAI is working on, and it could be unveiled this year before its 2027 release.

Of course, this all depends on the outcome of the Apple lawsuit against OpenAI, which alleges the AI lab illegally campaigned to obtain Apple trade secrets it could use to build its own hardware devices. In the lawsuit, Apple is requesting that OpenAI stop using Apple’s trade secrets or disclosing them, return stolen hardware, pay damages, destroy illegally obtained information, and, most importantly, guarantee that OpenAI products are not based on Apple’s plans and designs, which may be what ultimately delays or halts OpenAI’s hardware roadmap.

The fact that OpenAI's first hardware device could be a smart home speaker rather than a wearable is surprising. An AI wearable, such as a smart pin or smart glasses, seemed like a natural choice given the rapid rise in demand for smart glasses over the past year and the natural fit that wearables would be for adding more context to AI. However, no one has yet been able to popularize a pin or smart glasses for mainstream adoption. Smart home devices, such as Echo and Google Nest products, have already become mainstream, making it a less risky category for OpenAI. Its biggest challenge, though, will be convincing people they can trust it with their home data, especially given the sensors and cameras involved. This backlash is already showing up in reactions online. The cynical point of view would also question the timing of this report, coming right after the announcement of Apple's lawsuit. It's hard not to wonder if OpenAI had been testing multiple devices and decided to push the smart speaker as its first device since it would be less legally fraught than an AI wearable that could be influenced by unreleased Apple products via former Apple employees now at OpenAI.