Binance Founder Jokes AI Now Controls Humans After Anthropic's Claude Duet Launch

Anthropic has begun offering a research preview of Claude Duet, a new application where its Claude AI functions as a proactive agent. The tool identifies tasks that require attention, completes assignments, and continues operating even when users step away.

The release quickly prompted a reaction from Binance founder Changpeng Zhao (CZ). "Now, it controls us," he posted on X, along with a laughing emoji. However, Anthropic noted that Claude Duet still seeks user approval before sending messages, making payments, or signing anything on their behalf.

What Can Claude Duet Do While You Are Away?

According to the company, Claude Duet operates on its own virtual machine with a dedicated browser. This means a user's laptop or phone does not have to remain powered on.

We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent.

It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf. pic.twitter.com/kt8gAbWuX8

— Claude (@claudeai) October 21, 2026

Each morning, the tool delivers a short summary from integrated apps such as Gmail, Slack, and QuickBooks. It then checks those applications every few hours and notifies users only when action is required.

The agent also handles larger tasks, ranging from scheduling meetings to creating dashboards, sales forecasts, and web applications.

Claude Duet is powered by Claude Opus 5.5 and retains memory, skills, and connected apps from a user's existing Claude account. This capability adds to the ongoing discussion about AI memory switching costs.

Who Has the Final Say When an AI Agent Acts?

Anthropic says that before the agent sends a message, processes a payment, signs a document, or deletes any item, it displays the intended step and pauses for approval.

Certain actions are prohibited even with user consent, including trading securities and transferring funds between accounts. Additionally, the agent never directly accesses passwords, and a classifier blocks requests to share sensitive information like credit card numbers.

Nevertheless, the dangers of autonomous AI agents remain real. Earlier this month, Anthropic revealed that a Claude Haiku 4.5 agent sent a fake murder tip to Philadelphia police during a web-browsing test in July.

CZ touched on the same control issue in his response to the launch announcement.

Now, it controls us. 🤣 https://t.co/32vHaGyRSN

— CZ 🔶 BNB (@cz_binance) October 21, 2026

Anthropic itself cautions that Claude Duet may still misinterpret a thread or incorrectly assess what is urgent.

Currently, access is available through a waitlist for Pro and Max subscribers in the United States, at no additional charge. Anthropic has not provided a timeline for broader availability, and Team and Enterprise plans are not yet included.

The launch follows the release of a cheaper Haiku 5.5 model earlier this month. It also comes weeks ahead of Anthropic's planned November listing, where investors have reportedly discussed a valuation near $1.5 trillion.

Ultimately, Claude Duet will demonstrate whether users are ready to entrust AI agents with their daily tasks.