Cryptocurrency exchange Bybit EU has released an open-source tool that allows popular artificial intelligence assistants like ChatGPT and Claude to trade digital assets directly on behalf of users, bringing autonomous machine trading to everyday investors.
By Aisha Okonkwo | November 9, 2026
For regular crypto holders, this release marks a noticeable shift in how personal portfolios can be managed. Instead of spending hours tracking price charts, calculating position sizes, or navigating dense exchange order books, investors can now instruct an AI assistant using plain English sentences to inspect balances, rebalance crypto holdings, or execute buy and sell orders. However, handing real trading authority to conversational software also introduces fresh portfolio risks if a chatbot misinterprets a command during volatile market swings.
The Announcement
On November 9, 2026, during the opening day of the Web Summit 2026 technology conference at the MEO Arena in Lisbon, Portugal, Bybit EU announced the public release of the Bybit EU AI Trading Skill. Bybit EU operates out of Vienna, Austria, as the European division of global exchange Bybit, functioning under the European Union’s Markets in Crypto-Assets (MiCAR) regulatory framework.
The release comes as Bybit establishes a significant presence at the Lisbon conference, which runs from November 9 through November 12, 2026. Bybit Global Chief Technology Officer Rockman Zhang is scheduled to address conference developers on re-engineering high-throughput systems capable of handling more than 1.5 trillion USD in trading volume for the AI era. Concurrently, Bybit EU Co-Chief Executive Officer Georg Harer is participating in stage sessions on digital asset regulation, institutional adoption, and European compliance.
To showcase the new software integration, the exchange is hosting an interactive masterclass on November 10, 2026, titled “Agents on the Order Book: When AI trades crypto.” The session will feature live demonstrations of autonomous software agents connecting directly to central limit order books, testing how well machine models navigate live execution conditions.
What It Actually Does
The newly launched integration bridges the gap between conversational AI assistants and active cryptocurrency exchange accounts. Rather than forcing users to code custom automated scripts or manage complex developer tools, the tool packages exchange commands into a standardized set of instructions that compatible AI models can interpret instantly.
The code is hosted publicly as an open-source repository on GitHub under the directory BYBIT-EU/skills. Once connected to an account, the tool gives AI assistants access to several key capabilities:
- Zero-installation setup — Users can activate trading capabilities simply by directing their AI assistant to fetch the open instruction file from GitHub, removing the need for terminal commands or custom software downloads.
- Natural language order placement — Traders can execute spot market buys or set limit orders across listed crypto pairs by typing standard sentences such as buying a set dollar amount of an asset at a specified price.
- Spot-margin trading support — Advanced participants can access spot-margin trades with leverage of up to 10x, borrowing capital directly through chat commands.
- Automated portfolio tracking — The assistant can query account balances in real time, calculate asset allocation breakdowns, and recommend portfolio rebalancing adjustments.
- Broad model compatibility — The framework is built to operate across major conversational and coding assistants, including ChatGPT, Claude, Google Gemini, Cursor, and Codex.
By standardizing these communication rules, the project effectively allows everyday conversational models to function as interactive, automated trading terminals.
How It Works, in Plain English
To understand how this AI trading skill works, imagine dining at a busy restaurant. Today, trading crypto on an exchange is like having to walk back into the kitchen yourself. You have to find the chef, measure out the ingredients, check the stove temperature, and supervise the cooking. If you press the wrong button on a confusing dashboard or miscalculate an order type, you end up with a ruined meal and lost money.
With this integration, your AI chatbot acts as your personal waiter. You sit at your table and say: “Please take 200 USD from my balance and buy Ethereum if the price dips.” The chatbot takes your request, walks over to the exchange kitchen using a secure digital connector known as an API (Application Programming Interface), places the order with the chef, and returns to your chat window to confirm that your order was filled.
Security rules are built directly into this process. Users create a private access key on the exchange platform that is restricted strictly to Read and Trade permissions. Withdrawal permissions are strictly disabled, ensuring that even if an AI assistant makes an error, it cannot transfer funds out of your account to an outside wallet. I checked the open-source repository on GitHub myself: the entire skill configuration is a single setup file containing instructions and API routing rules without hidden compiled binaries. This transparent layout ensures that traders can inspect exactly how their assistant will talk to the exchange before sharing their restricted credentials.
Who Benefits
This integration offers clear advantages for everyday retail investors who find professional trading screens intimidating. Exchange terminals are frequently crowded with complex candlestick charts, order books, and unfamiliar order parameters. Being able to type instructions in conversational language removes technical friction for casual investors wanting to manage their own holdings.
Part-time crypto investors also gain a practical tool. Crypto markets operate around the clock, often catching working professionals or sleeping traders off guard during unexpected price swings. With Bitcoin trading at 74,650 USD and Ethereum at 2,119 USD, sudden market movements can rapidly alter an investment thesis. An AI assistant capable of tracking price milestones and executing planned trades helps investors maintain disciplined strategies without staying glued to their phone screens.
Additionally, European traders benefit from the regulatory structure backing the rollout. Because Bybit EU operates from Vienna under the European Union’s MiCAR framework, European users receive automated trading features within a supervised environment. With broader market sentiment currently registering in Fear at 29 on the Crypto Fear & Greed Index, using tools that follow clear compliance standards provides helpful reassurance compared to deploying unverified, offshore trading scripts.
The Open Questions
While conversational trading delivers major convenience, it introduces serious questions surrounding software reliability and financial risk. The most pressing concern is prompt hallucination and misinterpretation. Language models are probabilistic software programs, not infallible financial systems. If a chatbot misunderstands a user’s typed prompt, confuses similar token tickers, or executes an inverted trade during high volatility, the user bears complete responsibility for the financial outcome.
The inclusion of 10x leverage on spot-margin trading amplifies this risk significantly. Leverage represents borrowed money that magnifies both market gains and losses. Under 10x leverage, an adverse price move of just 10 percent can trigger a liquidation, wiping out an investor’s entire collateral. Entrusting margin debt to conversational prompts creates severe downside potential for everyday users who may not realize how fast leveraged positions can unravel.
Centralization issues also remain unresolved. While the skill configuration is open-source, the underlying AI models run inside centralized cloud data centers operated by major tech corporations. Should server latency, API outages, or rate limits occur during an abrupt market sell-off, automated instructions could fail to execute when investors need them most. European regulators under MiCAR will also likely monitor whether fully conversational order placement meets retail consumer protection standards, or whether future updates will require explicit confirmation prompts before orders reach an active order book.
Aisha Okonkwo’s take
Allowing a general-purpose language model to execute live orders bridges conversational software with real capital, but my main worry centers on automated prompt misinterpretation. Tracking the November 10 live demonstrations in Lisbon will show me whether these assistants handle market slippage cleanly under real trading conditions. If early users suffer erroneous leverage liquidations triggered by basic chatbot hallucinations, my assessment shifts toward keeping autonomous tools strictly confined to read-only market research.
The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.
10x leverage through a chat prompt is genuinely insane. one confused ticker during a wick and your collateral is gone before you finish typing the correction
counterpoint: if you typed short when you meant long thats on you. the bot just follows bad instructions faithfully
checked the BYBIT-EU/skills repo, its literally one setup file and withdrawal perms are disabled by default. still not letting claude near my limit orders during a dump tho
^ the withdrawal lock is the only reason id even test this. hallucinated execution risk still sits with you though
MiCAR supervision plus read and trade only keys makes this way less sketchy than the offshore bots people already run. watching the nov 10 Lisbon demo for the slippage test