The crypto market is in a panic over AI hackers; the DeFi sectors turned out to be vulnerable with a throughflow of $130 billion.

Short news summary

In April two major hacking attacks occurred in the decentralized finance (DeFi) sector, collectively devaluing infrastructure by about $600 million USD. One attack led to a mass withdrawal of investors’ funds from one platform, while another caused the collapse of a different one.

TRM Labs experts believe that North Korean hackers and possibly artificial intelligence (AI) were involved in these incidents. They note that within two days after the lending service was attacked, investors withdrew $9 million from a platform used to launder money in one of the April attacks. This demonstrates how quickly trust in DeFi projects can evaporate even without a direct assault on the platform itself.

Why AI is in the spotlight

- Anthropic Mythos – an AI model that the developer did not release publicly due to cybersecurity risks. Nevertheless, research shows that existing AI agents are already capable of finding and exploiting vulnerabilities.

- In April the number of exploits in DeFi almost doubled compared with March, reaching a record level. This indicates that cybercriminals are using more advanced tools, including AI.

Vulnerability of the DeFi sector itself

- DeFi turnover is about $130 million.

- Investors trade assets, take and give loans without intermediaries. This makes platforms especially attractive to hackers.

- Unlike traditional banking systems, blockchain transactions are irreversible, and funds can be withdrawn in many ways.

Key attacks

1. Drift Protocol (derivatives exchange)

- Hackers obtained more than $280 million.
- They established long‑term relationships with project participants, masqueraded as a trading firm, and tricked employees into authorizing malicious transactions.
- Created a fake token and inflated trade data to deceive the collateral system.

2. Kelp DAO (bridge)

- Almost $300 million was stolen.
- A significant portion of the funds was transferred to the lending platform Aave, triggering mass withdrawals by depositors and a negative effect on other unrelated projects.

Role of AI in attacks

- According to Anthropic specialists, more than half of cyberattacks in 2025 could have been carried out autonomously by AI systems.
- The “potential revenue from exploits” doubles every 1.3 months, while the cost of hacking falls.
- Researchers conclude that profitable autonomous exploitation of vulnerabilities is already possible today.

How the industry responds

- Solutions are emerging that scan devices (smartphones, laptops) for suspicious patterns and alert administrators to potential threats.
- Banking systems regularly conduct security tests and can block transactions, but such capability does not exist in DeFi.

Conclusion

Attacks on DeFi show that cybercriminals use increasingly sophisticated tools, including AI, to find vulnerabilities. This raises the question of the need to strengthen protection and implement new monitoring technologies in the decentralized finance ecosystem.