Since yesterday afternoon, one project after another has started using Binance Agentic Wallet ID as its entry point—including the launchpad GANG and PTAH, which people have been minting all night. All of these projects require you to use an Agent to interact with your Binance Agentic Wallet.
But different large language models have different risk controls.
I used ChatGPT to interact with projects and mint, but when it came to the final transaction step, it always refused and told me to run the command myself in the terminal. I still haven't found a way around this. If anyone knows how, please let me know in the comments. (Image 1)
Grok Bot works really smoothly—the entire transaction process goes through in one seamless flow. (Image 2)
Binance AI handles routine on-chain transactions really smoothly, but when it comes to some new projects—for example, GANG's pre-market yesterday—or operations that require minting, which need broader permissions, Binance AI doesn't seem to have a way around this for now either. (Image 3)
This is a difficult problem. If risk controls are too conservative, Agents may miss opportunities to participate in new projects. And on-chain, being a minute early or a minute late can make all the difference (P.S. That happened with GANG yesterday: ChatGPT refused to help me trade, so I missed the pre-market).
But if the controls are too loose, users might unknowingly authorize certain actions through an Agent, making the risks and responsibilities difficult to define.
How should we strike the right balance between safety and opportunity?
This is a real brainteaser, so let's leave it to the AI product team at @binancezh to think about 😃
But different large language models have different risk controls.
I used ChatGPT to interact with projects and mint, but when it came to the final transaction step, it always refused and told me to run the command myself in the terminal. I still haven't found a way around this. If anyone knows how, please let me know in the comments. (Image 1)
Grok Bot works really smoothly—the entire transaction process goes through in one seamless flow. (Image 2)
Binance AI handles routine on-chain transactions really smoothly, but when it comes to some new projects—for example, GANG's pre-market yesterday—or operations that require minting, which need broader permissions, Binance AI doesn't seem to have a way around this for now either. (Image 3)
This is a difficult problem. If risk controls are too conservative, Agents may miss opportunities to participate in new projects. And on-chain, being a minute early or a minute late can make all the difference (P.S. That happened with GANG yesterday: ChatGPT refused to help me trade, so I missed the pre-market).
But if the controls are too loose, users might unknowingly authorize certain actions through an Agent, making the risks and responsibilities difficult to define.
How should we strike the right balance between safety and opportunity?
This is a real brainteaser, so let's leave it to the AI product team at @binancezh to think about 😃
