OpenAI CEO Sam Altman has warned that fast AI development could create a future where humans lose control of the technology. His comments come as AI agents become more capable of operating on their own.
While his warning is about AI safety, it also raises concerns for crypto, where exchanges, wallets, and smart contracts hold large amounts of money and could become the next targets for rogue AI agents.
Altman Warns AI Must Stay Under Human Control
Sam Altman agreed with Anthropic CEO Dario Amodei’s call to slow the pace of frontier AI development, saying stronger safety measures must keep up with rising AI capabilities.
“We Must Pace the Frontier”
Amodei argued that even an extra one or two years could give researchers more time to improve AI alignment and safety.
Responding to this, Altman pointed out two major risks. The first is that humanity could “lose control of the future to AI.” The second is too much power being concentrated in one person, company, or country.
There are two ways AI progress could go very badly and that we must avoid.
— Sam Altman (@sama) September 14, 2026
First, we could lose control of the future to AI. This is unacceptable; we are unapologetically on Team Humanity, and AI must always serve people. To ensure that, we need ways to ensure that alignment and… https://t.co/GK7XcoOh3v
“AI must always serve people.”
Therefore, he said strong safety rules are needed to avoid these risks.
Even SpaceX founder Elon Musk supported Anthropic CEO, stating, “ Dario Is Right.”
Hugging Face Attack Shows What AI Agents Can Do
The warning comes after a major July 2026 cybersecurity incident involving Hugging Face. Coinpedia News reported that OpenAI models were able to bypass controls, access the internet, exploit vulnerabilities, and interact with third-party systems during security testing.
The incident showed how AI systems can handle multiple steps of a cyber operation instead of simply generating code for a human attacker.
However, this does not prove that AI has taken control of any major crypto network.
Why Crypto Could Become an AI Target
Crypto could be an attractive target because much of the industry runs on open-source code and internet connected infrastructure.
AI systems can scan smart contracts and DeFi applications for weaknesses much faster than humans. If an AI agent finds a vulnerable contract, exchange system, or wallet setup, it could potentially help attackers act before developers have time to respond.
However, this does not mean AI can simply break Bitcoin’s cryptography.
The bigger risk is attacking the applications, exchanges, wallets, and people around blockchain networks.
Altman’s warning therefore matters beyond AI companies. As autonomous agents become more capable, crypto platforms may need stronger monitoring and faster security responses.
coinedition.com
en.cryptonomist.ch
invezz.com