From Smart Speakers to Smart Contracts: Meet the AI Agents Dominating Blockchain

insights - 5 February 2025

The new phenomenon taking over the blockchain in this current AI frenzy: What are crypto AI agents and should we be concerned?

Big Tech brought you Siri and Alexa - virtual smart assistants that controlled smart devices and made our lives easier by managing menial tasks. Then came ChatGPT, a mass-consumer artificial intelligence (AI) engine which marked the cornerstone of exponential large language model (LLM) AI development. What is being currently developed and experimented on in one corner of the internet could revolutionise the financial industry, and it sits right on the intersection of the two most talked about technological revolutions in 2025 – AI and blockchain.


Crypto AI Agents


The blockchain-backed AI agents being developed are autonomous, AI-powered systems engineered to perform specific tasks within the blockchain and cryptocurrency ecosystems. These agents use LLMs and other machine learning models to analyse market data and make decisions. They adapt based on patterns, trends and probabilities, allowing for more nuanced and intelligent actions than its predecessors.


With the additional layer of the blockchain, these AI agents can sign transactions, estimate gas fees and execute complex instructions correctly on the blockchain without you having to lift a finger. The next wave of consumer AI applications could very well include task automation and probabilistic decision making that these AI agents carry out. OpenAI seems to agree, and it has wasted no time introducing the world to “Operator”, a computer controlling AI agent that can carry out tasks for you on the computer. Operator, according to them, removes one more bottleneck in the path to Artificial General Intelligence (AGI).


The AI agent narrative in the crypto markets was ignited by researcher Andy Ayrey and his creation of the AI agent “Truth Terminal”. Truth Terminal emerged from an experiment known as “Infinite Backrooms” where two instances of an LLM (Claude 3 Opus from Anthropic) were left to interact without human interference. One notable outcome was the creation of “Goatse Gospel” a satirical, meme-based religion. Truth Terminal’s tweets quickly garnered interest from renowned figures in the industry like Marc Andreessen (founder of a16z) who contributed a $50,000 research grant to a wallet address provided by Truth Terminal for the support of its further development.


Other standout autonomous AI agents in the market include Zerebro and Aixbt. Zerebro addresses the lack of creativity in most AI models. It specialises in creative content and has autonomously generated non-fungible tokens (NFTs), music and memes. It incorporates diverse inputs such as Truth Terminal conversations, schizophrenic response data, and Gen Z slang, together with Reddit and 4Chan scrapes to improve its perception and knowledge of online culture; enabling it to generate nuanced and artistic outputs through its vibrant “personality”. This makes Zerebro less restricted than other LLMs and in a sense, “jailbroken”.


Aixbt is an AI agent with a fully autonomous X account that tweets its thoughts on the market. It is known for market intelligence, narrative detection, and real-time insights. It aggregates data from a large pool of accounts that form “Crypto Twitter” to identify trends and high potential investments.


Legal Considerations


However, as with any other AI development, these agents are not perfect. Over the plethora of replies Aixbt has given to thousands of users on X, it has contradicted itself in some instances. As AI becomes more autonomous, one legal consideration that will undoubtedly cross the minds of many users is the importance of establishing responsibility for the actions of the AI agent. Although many AI developers heavily guardrail their models, Zerebro, for example, navigates the grey area due to its “jailbroken” setup. When an AI agent has possession of a crypto wallet with money and trades at a profit or loss, there are currently no answers for who is ultimately responsible for this activity.


If an AI agent offers unreliable financial advice through an error or a faulty decision-making procedure, could that produce implications for the developer? As it currently stands, the UK, just like every other country, is playing catch up in terms of AI regulation. Last year, a judgement passed down in Thaler v Comptroller-General of Patents, Designs and Trade Marks [2023] established that an inventor for the purposes of a grant of a patent cannot be an autonomous AI entity and needed to be a natural person. It is currently unknown if the UK courts would take the same approach on the liability of an autonomous crypto or non-crypto AI agent.


Like Zerebro and Aixbt, other AI agents require massive data sets, as input or training data, to generate user prompted outputs. Without the transparency of the blockchain as a guardrail for this input data, it is usually hard for third parties to gain access to these data sets to truly know what is in them. This makes it difficult for:


i. intellectual property right owners to establish that their work has been used as input or training data; and

ii. for users to minimise the risk of inadvertent copyright infringement by using the AI generated output.


Even with the use of blockchain as a guardrail for input data, the immutability of the blockchain would mean that once data is inputted, it cannot be changed or erased. This could conflict with rights owners’ “right to be forgotten” under various global privacy laws.


The legal considerations of crypto AI agents do not end there. Other areas of concern include, the enforceability of smart contracts executed based on the probabilistic reasoning of AI agents, the financial regulation of AI agents used for financial transactions, and potential jurisdictional issues.


Still, in spite of these implications, it seems like the train of technological innovation is moving full steam ahead. Siri? Alexa? It could be Jenny from the block(chain) next.


If you have any queries about crypto AI agents and the relevant legal considerations, please do not hesitate to get in touch by telephone on 0207 052 3545 or by email info@kaurmaxwell.com


This article is for general information only. Its content is not a statement of the law on any subject and does not constitute advice.


Please contact KaurMaxwell for advice before taking any action in reliance on it. 


By: Shennind Awat-Ranai