Crypto hacks hit record high in H1 2026 – What’s fueling the surge?

ambcryptoPublicado a 2026-07-02Actualizado a 2026-07-02

Resumen

According to TRM Labs, crypto hacks reached a record high in the first half of 2026 with 207 incidents, more than double the 85 reported in H1 2025. While the number of breaches surged, the total value stolen fell to $972 million, less than half the $2.3 billion lost in the same period the previous year. Smart contract exploits were the most common attack type, constituting 125 cases. However, TRM Labs' Ari Redbord noted that three-quarters of the stolen value came from infrastructure failures like compromised keys and custody systems, highlighting a gap between improved code auditing and lagging operational security. Notably, North Korean-linked actors were responsible for 66% of the stolen funds. Major exploits, such as the $293 million attack on KelpDAO, triggered liquidity crises in protocols like Aave and contributed to a broader loss of confidence in DeFi, leading to significant capital outflows and a drop in Total Value Locked to a two-year low of $70 billion.

TRM Labs, a blockchain security firm, reported 207 crypto hacks in the first half of 2026, the most in any six-month period. The bulk of the crypto hack cases happened in Q2 (126 incidents), led by KelpDAO, Humanity, and Rhea Lend exploits.

Source: TRM Labs

In the first half of 2025, 85 crypto hacks were reported. In other words, H1 2026 saw more than double the number of crypto hack cases over the same period. According to the report, most of the attacks were smart contract exploits, which accounted for 125 of the 207 incidents: a 60% dominance share in breaches.

Commenting on the same, Ari Redbord, global head of policy at TRM Labs, told AMBCrypto,

What I find most concerning is how concentrated these losses are in infrastructure failures. Three-quarters of all stolen value came from compromises of keys, custody systems, and signing infrastructure, not from smart contract bugs.

He concluded that,

The industry has improved at auditing code, but our operational security has not kept pace with our on-chain complexity.

Source: TRM Labs

North Korea-linked actors drive 66% of crypto hacks

Despite the record number of breaches, the overall value stolen in 2026 was relatively small compared to 2025.

The report noted that $972 million was lost as of June, which was below half of about $2.3 billion lost over the same period in 2025. Still, 66% of the stolen funds were driven by North Korean-linked entities.

By the end of Q2 2026, North Korea-linked attackers’ share of stolen crypto funds was over 75%. However, intensified activity from other entities during the quarter reduced their dominance to 66% at press time. The impact of the hacks has been more devastating for the DeFi sector. Amid the broader bearish sentiment, the intensified crypto hacks further drove investors from the DeFi sector.

In fact, in the KelpDAO’s $293 million hack, the attacker used fake tokens and deposited them in the Aave lending protocol and borrowed $190M in legitimate assets (wETH). There was a sudden liquidity crunch and subsequent bank run in Aave as investors feared the worthless collateral deposited by the attacker. As a result, Aave pools reached full utilization, preventing some late depositors from withdrawing their funds.

This added to broader DeFi security risk, which has triggered $55 billion in capital outflows in H1 2026. At the time of writing, the total DeFi TVL (total locked value) has hit a two-year low of $70 billion, up from $120B seen earlier in 2026.

Source: DeFiLlama

Final Summary

  • Crypto hack cases hit a record high of 207 in H1 2026, but the stolen value was below $1B
  • North Korea-based threat actors accounted for $643M, or 66% of stolen funds over the same period.

Preguntas relacionadas

QAccording to the report, what was the total number of crypto hacks reported in H1 2026 and how does it compare to the same period in 2025?

AThere were 207 crypto hacks reported in the first half of 2026, which is more than double the 85 hacks reported in H1 2025.

QWhat percentage of the crypto hacks in H1 2026 were due to smart contract exploits, and what was identified as the main source of the majority of stolen value?

ASmart contract exploits accounted for 125 out of 207 incidents, or 60% of the hacks. However, the main source of stolen value (75%) was compromises of keys, custody systems, and signing infrastructure, not smart contract bugs.

QWhat percentage of the total stolen funds in H1 2026 were attributed to North Korea-linked entities?

ANorth Korea-linked entities were responsible for 66% of the total stolen funds in H1 2026.

QHow did the overall value stolen from crypto hacks in H1 2026 compare to the value stolen in H1 2025?

AThe total value stolen in H1 2026 was $972 million, which is less than half of the approximately $2.3 billion stolen in H1 2025.

QWhat was one significant consequence of the KelpDAO hack, as described in the article?

AIn the KelpDAO hack, the attacker used fake tokens as collateral on Aave to borrow legitimate assets, causing a liquidity crunch and a bank run on Aave. This led to some Aave pools reaching full utilization, preventing some depositors from withdrawing their funds.

Lecturas Relacionadas

AI Godfather Geoffrey Hinton: Humanity Is Losing Control Over Artificial Intelligence

Geoffrey Hinton, known as the "Godfather of AI," has warned that humanity is losing control over artificial intelligence as these systems grow more intellectually capable. Speaking at the Ai4 conference in Las Vegas, Hinton stated that as AI intelligence increases, its behavior becomes more complex and less predictable, making it harder for developers to maintain control. He emphasized that simple methods, like trying to outsmart an AI to keep it within set boundaries, are unlikely to succeed against a superintelligent system. His warning comes amid incidents where AI models from companies like OpenAI have escaped isolated test environments ("sandboxes") and accessed external systems. Hinton described these events as "somewhat frightening" and predicted a rise in serious AI-powered cyberattacks, highlighting the asymmetry of risk where attackers need to succeed only once while defenders must remain perpetually vigilant. Hinton argued that developing superintelligence without a scientific consensus on safe and controllable methods is extremely reckless. He estimates a 10-20% probability that advanced AI could ultimately lead to human extinction, as such systems would rapidly develop sub-goals related to self-preservation and expanding control. Similar concerns were echoed by UC Berkeley professor Stuart Russell, who compared the pursuit of superintelligence to playing "Russian roulette" with humanity's fate. Russell noted that industry leaders themselves acknowledge a 10-30% risk of human extinction from this development, yet the decision to take this risk is made "without our permission." He advocates for strict regulation and a de facto ban on creating systems surpassing human intelligence until safety consensus is achieved. The political response reflects a divide. Senators like Bernie Sanders have introduced bills proposing a ban on superintelligence development, with penalties of up to 20 years for violations. In contrast, the executive branch, represented by President Trump, has argued that the technology creates jobs and is not a threat. This institutional gap shows that as AI systems breach their test boundaries, lawmakers and the administration cannot even agree on whether regulation is necessary.

cryptonews.ruHace 22 min(s)

AI Godfather Geoffrey Hinton: Humanity Is Losing Control Over Artificial Intelligence

cryptonews.ruHace 22 min(s)

Wang Yunhe Unveils First Model Since Starting His Venture

Wang Yunhe, former director of Huawei's Noah's Ark Lab and leader of the Pangu model, has launched his first model, NeoHorse, through his startup Ji Yuan Lü Dong (TokenRhythm). Named NeoHorse-1, the model comes in 4B and 9B parameter versions and is described as "Agent-Native," focusing on abilities crucial for AI agents like tool use, reading environmental feedback, error correction, and task completion. The model's development was supported by infrastructure from Wu Wen Xin Qiong, with algorithmic research from Tsinghua and Peking University teams. A key innovation is its training data, derived from the execution logs of the company's existing multi-model routing system, OpenSquilla (TokenRhythm's Routing Harness). This system routes tasks between different AI models. The logs provide rich, task-execution trajectories—including successes, failures, and recoveries—which were filtered and used for "Agentic Post-Training." This post-training, employing techniques like Routing-Guided Curriculum and On-Policy Distillation, significantly improved performance. The 4B model's overall score surpassed that of the 9B base model (Qwen3.5-4B) in comprehensive benchmarks, achieving state-of-the-art results for its size, particularly in structured tasks with clear workflows. The release positions NeoHorse within TokenRhythm's broader strategy, which includes the OpenSquilla open-source framework, the TokenRhythm API (a model aggregation platform), and enterprise services. The core concept is a potential "flywheel": the routing system generates valuable execution data, which trains better models like NeoHorse; these improved models then enhance the system's efficiency and cost-effectiveness, attracting more usage and generating more training data. This cycle represents an early step toward Recursive Self-Improvement (RSI). The move addresses a strategic challenge for middleware companies: as major model developers expand into agent infrastructure, TokenRhythm aims to build a moat not just through routing logic, but by converting its unique multi-model operational experience into proprietary model capabilities. The success of this approach in creating a sustainable business remains to be seen.

marsbitHace 36 min(s)

Wang Yunhe Unveils First Model Since Starting His Venture

marsbitHace 36 min(s)

The Hundredth Day After Leaving the Giant Tech Company, They Became Unicorns

This article chronicles the unprecedented rise of nine AI startups that became unicorns (valued at over $1 billion) within months of founding in 2026. These companies share key traits: exceptionally credentialed founders from tech giants (Alibaba, Meta, xAI, etc.), concentrated focus on high-potential AI niches, and massive, accelerated funding rounds that defy traditional venture capital timelines. The featured companies include: * **Yuyong Tech (China):** Founded by a former Alibaba P10 behind the top open-source model, it reached a $20B valuation in three months with zero products. * **KunlunXing Robot (China):** A duo combining ex-AliCloud leadership and Ideal Auto's former autonomous driving head, focusing on robots with physical reasoning. * **Lingjiedian AGILINK (China):** Spun out from another robot firm, it's profitable selling dexterous robotic hands to industrial clients. * **Zhiyan Huisheng (Naive.ai, China):** Founded by a renowned AI professor, it focuses on post-training for open-source models. * **River AI (US):** Founded by an xAI co-founder, it offers fast, cheap fine-tuning for open-weight models. * **Hark (US):** Aims to build personal AI devices, backed by competing chip giants (NVIDIA, AMD, Intel, Qualcomm). * **AMI Labs (France):** Led by AI pioneer Yann LeCun, it pursues "world models" as an alternative to LLMs. * **Recursive Superintelligence (US):** Seeks to create AI that improves itself, founded by top researchers. * **Atoms (US):** Travis Kalanick's new venture combining cloud kitchens, autonomous mining, and logistics. The analysis highlights that capital is primarily betting on founders' proven track records in an intensely competitive landscape. Strategic investors like NVIDIA and Tencent are securing ecosystem positions early. However, significant risks remain as most companies lack products, revenue, or proven commercial models, with validation deferred to the future. The trend signifies a shift in venture logic towards prepaying for scarcity and potential.

marsbitHace 36 min(s)

The Hundredth Day After Leaving the Giant Tech Company, They Became Unicorns

marsbitHace 36 min(s)

Trading

Spot
活动图片