The online crypto community has found a new chatbot powered by artificial intelligence (AI) that can either warn developers about flaws in smart contracts or teach hackers how to use them.
ChatGPT, a chatbot tool made by the AI research company OpenAI, came out on November 30. According to the company, it was made to talk “in a conversational way” and be able to answer follow-up questions and even admit mistakes.
But some Twitter users have realized that the bot could be used for both good and bad since it can be told to find holes in smart contracts.
Related: Microsoft Power Platform Certification empowering Pakistani Young Girls to become software application developer
Stephen Tong, the co-founder of Zellic, a company that audits smart contracts, asked ChatGPT to help him find a flaw in some smart contract code.
The bot replied that the contract had a reentry vulnerability that allowed someone to take money out of the contract over and over again, and it gave an example of how to fix the problem.
In May, an attacker on the decentralized finance (DeFi) platform Fei Protocol used a similar exploit to steal $80 million.
Others have shared what the chatbot told them after they gave it weak smart contracts. A Twitter user named devtooligan posted a screenshot of ChatGPT, which gave the exact code needed to fix a Solidity smart contract vulnerability, with the comment “We’re all going to be out of a job.”
Twitter users have already started to joke that they can start auditing the security of businesses by using the bot to look for flaws in smart contracts.
Asian Trade tested ChatGPT and found that it can also create an example smart contract from a prompt using simple language, generating code that could apparently provide staking rewards for Ethereum-based nonfungible tokens (NFTs).
ChatGPT’s example of a smart contract made with Solidity for staking NFT rewards from a simple prompt Cointelegraph is a picture.
Even though the chatbot can test how smart contracts work, it wasn’t made just for that, and many people on Twitter have said that some of the smart contracts it makes have problems.
It might also give different answers depending on how it’s asked, so it’s not perfect.
Sam Altman, the CEO of OpenAI, said on Twitter that the tool was “an early demo” and that it was “very much a research release.”
He thought that “language interfaces are going to be a big deal” and that tools like ChatGPT will “soon” be able to answer questions and give advice. In later versions, these tools will be able to finish tasks or even learn new things.