Hong Kong Courts Gulf Capital as China Reins In AI Companion Apps

LEAP East, the Riyadh-founded technology conference, held its first Asia edition in Hong Kong this week, drawing more than 450 exhibitors, most of them from mainland China, Hong Kong and Saudi Arabia, with a smaller contingent from Europe. The lineup leaned toward biotech, robotics and software, and the now-familiar cast of dancing Chinese humanoid robots was on display, including one dressed in a shemagh, the traditional Gulf headscarf, to court visitors from the region.

The staging captured Hong Kong's wider strategy. Facing tighter Western scrutiny as geopolitical tensions climb, the city has spent recent years pursuing Gulf capital with increasing intent. The most concrete marker so far is a memorandum of understanding signed with Saudi Arabia's Public Investment Fund in late 2024, aimed at a joint $1 billion fund to back Hong Kong companies expanding into the kingdom.

What that engagement translates into on the ground is harder to measure. The actual volume of Middle Eastern investment reaching Hong Kong remains difficult to quantify, according to Nikkei Asia, a gap between announced intent and traceable flows that is worth keeping in view as more deals are unveiled.

Chinese firms move first

Chinese companies are not waiting on the diplomacy. Several are using Hong Kong as a staging ground before pushing into the Gulf and further afield. Keeta, the overseas brand of Meituan, is the clearest case. It has become Hong Kong's largest food delivery platform after displacing Deliveroo, and has since expanded into Saudi Arabia, the United Arab Emirates, Qatar, Kuwait, Bahrain and Brazil, with some backing from the Hong Kong government.

Entering those markets has required more than replicating the domestic model. Speaking at a LEAP East session, Keeta CEO Tony Qiu pointed to operational adjustments, such as accommodating delivery riders who may need to pause and pray during a route. That is a departure from Meituan's earlier domestic approach, under which riders faced fines for missing delivery windows, a policy Nikkei Asia notes has softened amid regulatory pressure in China.

Beijing pulls back AI companions

At home, the direction is toward tighter control. As authorities intensify oversight of the AI sector, China's largest tech companies are rolling back AI persona features, disappointing users who over the past year have turned to virtual characters for companionship and emotional support, particularly on ByteDance's Doubao chatbot, according to Nikkei Asia.

The rollback precedes new rules taking effect July 15 that restrict AI services from offering virtual intimate relationships, such as AI family members or romantic partners, to minors. The regulations also require parental consent for other anthropomorphic AI services aimed at children under 14, reflecting official concern over privacy risks, harmful content and possible psychological effects.

The move exposes the friction between the popularity of AI companionship and Beijing's push for stricter controls. The Cyberspace Administration of China has already pulled thousands of noncompliant AI products in a nationwide campaign targeting data security, misinformation and obscene content, Nikkei Asia reported.

Workarounds on Claude

On a separate front, Anthropic is moving to close loopholes that let Chinese companies get around its restrictions on unauthorized use in the country, according to reporting by the Financial Times. People familiar with the matter said Chinese firms including Ant Financial had reached Anthropic tools such as Claude Code through methods that involved cloud providers and overseas subsidiaries.

Ant provided employees with corporate Claude accounts accessed via the company's intranet, which connects to its Singapore-based entity, those people said. ByteDance does not facilitate access to Claude but this year introduced a reimbursement scheme letting engineers expense personal subscriptions, and those engineers use VPNs to reach them, according to five employees cited by the Financial Times.

The arrangements do not break U.S. or Chinese law but breach Anthropic's terms of service, which bar Chinese companies and foreign entities they own from using its models. Anthropic maintains one of the strictest bans among U.S. AI companies on usage in China, including user verification requirements.

Why it matters

The three threads point in the same direction: Chinese and Hong Kong companies are actively reshaping where and how they operate as access to Western markets and tools narrows. Hong Kong's Gulf outreach gives them a route outward, Beijing's persona rules redraw what is permissible at home, and the Anthropic dispute shows how far firms will go to keep access to leading U.S. models. Each carries an unresolved question, from the still-opaque size of Gulf inflows to how effectively either regulators or vendors can enforce the new limits.