Key Notes
- Vitalik Buterin argued that systems allowing greater human input produce safer and higher-quality results.
- He voiced support for open-weight AI models with strong editing capabilities and shared a vision for brain-computer interface technology.
- He compared human-in-the-loop AI, which offers oversight and error prevention, with fully autonomous AI, which delivers scale and creativity.
Ethereum ETH $4 297 24h volatility: 1.7% Market cap: $518.73 B Vol. 24h: $38.05 B co-founder Vitalik Buterin recently shared his opinion on the rampant development of “Agentic AI” models in the artificial intelligence (AI) industry.
In addition to his criticism, the Ethereum co-founder added that systems allowing greater human input produce better and safer results. This is the first major comment from Buterin, after his views on ETH Treasury firms last week.
Echoing something @karpathy recently said, it does frustrate me how a lot of AI development is trying to be as "agentic" as possible, when actually creating *more* paths for human input both creates a better output (now for quite a while going forward) and is better for safety.
— vitalik.eth (@VitalikButerin) August 11, 2025
Vitalik Buterin Slams Agentic Al Models
Ethereum co-founder Vitalik Buterin recently voiced concerns about the development of the artificial intelligence (AI) industry. He criticized the “Agentic AI models,” stating that the industry’s focus has shifted to bringing too much automation to systems.
In a recent post, Buterin agreed with comments by former Tesla AI director Andrej Karpathy, arguing that enabling more avenues for human input can both improve output quality and enhance safety.
He said he is more enthusiastic about open-weight AI models, which have strong editing capabilities.
Looking ahead, Buterin shared a vision for brain-computer interface (BCI) technology, which could display content as it is generated, measure his real-time reactions, and adjust accordingly.
In practical use-cases, human-input AI models and agentic AI models have their own set of benefits.
Human-in-the-loop AI offers greater output quality and oversight, allowing people to add context, identify errors, and set boundaries. This makes the system more reliable and less vulnerable to serious mistakes.
On the other hand, fully autonomous AI can operate at massive scale and speed, delivering creative solutions beyond human imagination or direct control.
However, removing human oversight significantly increases risk, as a single flawed output could spread quickly and widely.
Buterin Has a Message for ETH Treasury Companies
In another development, Buterin cautioned ETH Treasury firms racing to accumulate ETH coins, and avoid any sort of “over-leveraging.”
Bitmine Technologies and SharpLink Gaming have been in the race for strong ETH accumulation , with the former taking its ETH holdings over 1 million.
Earlier today, Bitmine reported its ETH holdings worth over $4.96 billion. The company holds 1,150,263 ETH, priced at $4,311 per token, positioning it as the largest Ethereum treasury globally.
next
