Ai-driven attacks push april crypto thefts over 634 million dollars

Rapid advances in artificial intelligence technology may have shifted the cybersecurity balance in favor of threat actors, according to leading industry voices. At the WAIB Summit in Monaco, Immunefi CEO Mitchell Amador pointed out that the latest generation of AI models has led to a resurgence of attacks within decentralized finance (DeFi).

Rising risk from ai-powered exploits

Amador emphasized that the growing accessibility of cutting-edge models like Claude Opus 4.8 and ChatGPT 5.5 has played a significant role in accelerating attacks on crypto asset platforms. He explained that these ai tools allow vulnerabilities to be identified more quickly, giving malicious actors a tactical edge in exploiting protocol weaknesses.

According to data from DefiLlama, industry-wide attacks spiked sharply in April 2026, with losses exceeding 634 million dollars from crypto platforms. This figure marked the highest monthly total since February 2025, when losses from the Bybit hack pushed cumulative damage to roughly 1.4 billion dollars.

Glossary: DefiLlama is an on-chain analytics platform that tracks total value locked, transaction volume, and attack statistics for decentralized finance protocols. It is a widely referenced source for comparative data across the DeFi ecosystem.

Critical years ahead for security

Discussing the sector’s future, Amador described the next three to four years as a crucial resilience test for crypto. During this period, he said, defense teams must develop robust codebases using ai tools to outpace attackers, enabling systems that malicious actors cannot easily penetrate.

He noted that this challenging phase could be shortened to less than two years if community-driven security solutions are more widely adopted. The analysis suggests that as security researchers become more proficient at deploying AI for defense, the current threat levels may begin to subside.

Renewed spotlight on DeFi attacks

Amador’s remarks came just after Anthropic introduced its new ai model, Claude Mythos (Fable 5), intensifying concerns within the industry about the potential for accelerating crypto hacks. Anthropic clarified that different security-related ai tasks were handled by a separate model, Claude Opus 4.8, which comes equipped with various safeguards.

These debates over cybersecurity have only become more urgent in the wake of a series of high-profile DeFi breaches. On April 19, an attacker withdrew around 116,500 restaked Ether from Kelp DAO’s bridge using LayerZero infrastructure. At the time, the seized funds were valued at between 290 million and 293 million dollars.

Glossary: DVN, or decentralized validator network, is a security layer for verifying cross-chain messages. Overreliance on a single validator or pathway in such networks can create a single point of failure.

LayerZero commented that Kelp DAO’s 1/1 DVN configuration relied solely on a single validator for cross-chain messages, introducing a critical vulnerability. The company noted that it had previously advised against such a setup. This incident has reignited discussion about how protocol design choices directly impact security outcomes.

The post Ai-driven attacks push april crypto thefts over 634 million dollars appeared first on COINTURK NEWS.