AI Agents Tested to Prevent $3.4B Crypto Hacks

TheNewsCryptoОпубликовано 2026-02-19Обновлено 2026-02-19

Введение

In 2025, the crypto industry suffered significant breaches, with three major incidents accounting for 70% of total thefts, including a $1.4 billion hack on Bybit. To combat vulnerabilities, OpenAI is testing AI agents in collaboration with Paradigm and OtterSec using EVMbench. These AI systems aim to detect and fix smart contract flaws more efficiently than human auditors, addressing the $3.4 billion lost to hackers in 2025. While AI agents excel at identifying vulnerabilities, concerns remain about potential misuse by attackers. Industry leaders suggest AI and crypto could become increasingly intertwined, with stablecoins and crypto payments playing key roles in future AI-driven systems.

In 2025, the crypto industry witnessed some massive breaches, with just three major incidents being responsible for around 70% of the overall money theft. It is noteworthy that the biggest hack is said to be the Bybit exchange hack, which alone swept around $1.4 billion.

For a solution, OpenAI is testing if AI agents can detect vulnerabilities in real blockchain spaces, working with Paradigm and OtterSec, using its EVMbench. Any bug associated with smart contract code will impact the money of big and small investors, as these automated programmes look after over $100 billion in open-source digital assets.

In 2025, hackers swept up around $3.4 billion from crypto platforms, and developers can now just feel how vulnerable the system is when hackers exploit weak code. Depending on human audits will not help anymore as live contracts face new and evolving attacks, and also, it takes a lot of time and costs.

Instead, AI agents can complete the same work in less time, so frameworks such as EVMbench are used by OpenAI. EVMbench leverages AI agents in test environments to aid developers in understanding how smart contracts may work under real-world pressure before the real deployment.

AI Agents are Performing Better

Primarily, the agents will find hidden vulnerabilities in code, fix the issue without breaking the function of the contract, and then attempt to exploit the weakness to drain funds if the problem still exists.

As per the prior results, AI agents are performing better at detecting vulnerabilities instead of fixing them safely. People are now worried that attackers could misuse AI-powered tools to detect weaknesses in blockchain systems more skilfully than ever.

Jermy Allaire, an American technologist, revealed that a lot of AI agents will leverage stablecoins to trade over blockchain networks. Changpeng Zhao, the founder and ex-CEO of Binance, also revealed that crypto could be the native payment layer for AI-driven systems in the future.

If this happens, then it will benefit both the users and the attackers as soon as they interact with contracts directly in real financial environments where actual money is at stake.

Highlighted Crypto News Today:

Hyperliquid Launches DeFi Lobby in Washington

TagsAIHackOpenAI

Связанные с этим вопросы

QWhat was the total amount stolen in crypto hacks in 2025, and what percentage did the three major incidents account for?

AIn 2025, hackers stole around $3.4 billion from crypto platforms, with the three major incidents accounting for approximately 70% of the total theft.

QWhich company is testing AI agents to detect vulnerabilities in blockchain, and what is the name of the benchmarking framework they are using?

AOpenAI is testing AI agents to detect vulnerabilities in real blockchain spaces, and they are using a framework called EVMbench.

QAccording to the article, what are the two main limitations of relying solely on human audits for smart contract security?

AHuman audits are limited because live contracts face new and evolving attacks that are hard to anticipate, and the auditing process is very time-consuming and costly.

QWhat specific tasks can AI agents perform to improve smart contract security, as mentioned in the article?

AAI agents can find hidden vulnerabilities in code, attempt to fix the issue without breaking the contract's function, and then test to see if they can still exploit the weakness to drain funds.

QWhat concern does the article raise about the potential misuse of AI-powered tools in the crypto space?

AThe article raises the concern that attackers could misuse AI-powered tools to more skillfully detect weaknesses in blockchain systems, posing a greater security risk.

Похожее

Anthropic Cries Wolf: Is the AGI Threat Real, or Just an IPO Story?

Anthropic has published an article titled "When AI builds itself," discussing the emerging concept of "recursive self-improvement," where AI begins to actively participate in designing, training, testing, and optimizing its own subsequent versions. The company presents internal data showing that by May 2026, over 80% of code merged into its codebase was written by Claude, its AI model. Claude's capabilities have expanded to handling complex, open-ended engineering tasks, achieving a 76% success rate in such areas, and even contributing to research processes, such as optimizing code performance and conducting AI safety experiments. Anthropic outlines an evolution from human-driven development to AI-assisted workflows, culminating in the current stage where AI agents can autonomously write, run, and delegate code. The company cautions that the path toward a "closed loop," where AI continuously improves itself, is becoming visible. It calls for coordinated global mechanisms to potentially slow or pause frontier AI development to allow safety research and societal structures to catch up. However, the timing of this warning coincides with Anthropic's preparations for an IPO, framing the narrative not just as a safety concern but also as a demonstration of Claude's advanced capabilities and its integral role in accelerating Anthropic's own R&D—creating a potential "flywheel" effect for competitive advantage. This contrasts with OpenAI's recent, more policy-oriented discussion of the same risks, highlighting the competitive dynamics in the AI industry as companies position themselves in both the technological and regulatory landscape.

marsbit55 мин. назад

Anthropic Cries Wolf: Is the AGI Threat Real, or Just an IPO Story?

marsbit55 мин. назад

Coinbase Targets Crypto Crime, Freezing $3M Linked To Scam Operations

Coinbase froze over $3 million in cryptocurrency linked to Southeast Asian scam networks as part of a multi-agency "Disruption Week" campaign. The operation, coordinated by the DOJ's Scam Center Strike Force, involved companies like Meta, Microsoft, and Starlink, and disrupted over 1.4 million scam accounts. The DOJ noted that investment fraud, including "pig butchering" scams, is among the fastest-growing threats, with crypto-related scams causing billions in losses. This action follows other recent crackdowns targeting scam infrastructure globally. Coinbase highlighted that blockchain provides a permanent transaction record, aiding investigations and countering the narrative that crypto is solely a tool for crime.

bitcoinist1 ч. назад

Coinbase Targets Crypto Crime, Freezing $3M Linked To Scam Operations

bitcoinist1 ч. назад

BIT Research: ETF Purchases Have Slowed, Strategy (MicroStrategy) Has Slowed, What Else Can Drive Bitcoin's Rise?

Market Refocus on Inflation and Rate Expectations Weighs on Bitcoin Currently, the market is in a phase of macro-repricing dominated by inflation and interest rate expectations. Bitcoin, which previously benefited from easy liquidity and low inflation, is seeing its core bullish drivers weaken. These drivers were market expectations for interest rate cuts and strong inflows from Bitcoin ETFs and institutions like MicroStrategy (referred to as "Strategy" in the text). The logic has shifted. Recent high inflation data (e.g., CPI hitting 3.8% in a May 2026 report) has caused the market to sharply reduce its rate cut expectations for 2025 and even price in potential hikes. This is a key constraint for Bitcoin, as it lacks cash flows and is highly sensitive to rate expectations. Concurrently, institutional capital flows have slowed significantly. Following the hot CPI data, Bitcoin ETFs saw accelerated outflows, with around $4.3 billion leaving over a period. MicroStrategy's ability to keep adding substantial Bitcoin to its balance sheet is also diminishing. Together, ETF and MicroStrategy holdings total roughly $110 billion, but their momentum as growth engines is cooling. In summary, Bitcoin's current pressure stems not from its own fundamentals but from a changing macro environment. As long as inflation stays elevated, Bitcoin is likely to remain in a consolidating phase. However, historically, inflation eventually peaks. Once it recedes and rate cut expectations rebuild, institutional capital could return, potentially fueling a new and more robust recovery phase for Bitcoin.

marsbit1 ч. назад

BIT Research: ETF Purchases Have Slowed, Strategy (MicroStrategy) Has Slowed, What Else Can Drive Bitcoin's Rise?

marsbit1 ч. назад

Earning 1000 Trillion in Half a Year, 'Pocketing' 20 Million per Capita: This Round of Wealth Creation in the Korean Stock Market is Unprecedented in Scale

The South Korean stock market is experiencing an unprecedented wealth surge in 2026, with household equity and fund asset values soaring by over 1,000 trillion KRW (~$730bn) year-to-date. This translates to an average per capita wealth increase of roughly 20 million KRW, fueled by a historic 109% rally in the KOSPI index. The boom is driven by three converging forces: an AI-driven semiconductor supercycle boosting giants like Samsung and SK Hynix; the government's "Value-Up" market reforms addressing long-standing corporate governance issues; and aggressive real estate regulations that have locked capital within financial markets, preventing profits from flowing back into property. This has triggered a wealth effect, boosting high-end consumption significantly. However, the gains are highly concentrated. The two semiconductor behemoths account for over half the index's value, but retail investors own relatively low stakes in them, systematically missing the biggest rallies. Wealth and consumption benefits are skewed towards luxury goods and imported cars, bypassing mainstream retail. Further risks stem from excessive leverage, with high trading volume in leveraged ETFs, and a market sentiment heavily reliant on the AI sector's fortunes and speculative rumors. While this cycle marks a potential shift from real estate to equities as a primary wealth generator for Koreans, its sustainability, amid structural imbalances and leverage, remains a critical test.

marsbit1 ч. назад

Earning 1000 Trillion in Half a Year, 'Pocketing' 20 Million per Capita: This Round of Wealth Creation in the Korean Stock Market is Unprecedented in Scale

marsbit1 ч. назад

Behind ZEC's Over 30% Plunge: An 'Unlimited Minting' Vulnerability with No Way to Prove if It Was Ever Exploited

A critical vulnerability was discovered in Zcash's Orchard privacy pool, allowing for the theoretical creation of undetectable counterfeit ZEC. Researcher Taylor Hornby found the flaw on May 29th, 2024, within the Orchard circuit's cryptographic constraints, which could let an attacker bypass asset conservation rules. Although a rapid emergency fix was deployed within days via a coordinated soft and hard fork, a core uncertainty remains: due to Orchard's privacy features, it is impossible to cryptographically prove whether this "unlimited mint" flaw was exploited in the nearly four years since the pool's 2022 launch. This uncertainty, rather than the patched flaw itself, triggered a market panic, causing ZEC's price to drop over 30%. While the Zcash Foundation stated no evidence of exploitation was found, independent entity Shielded Labs emphasized the impossibility of definitively proving no counterfeit ZEC was ever created. The incident highlights the unique trust challenge in privacy systems. To address this, developers are proposing a new network upgrade with enhanced auditing to allow verifiable proof of supply integrity. Notably, the researcher utilized the newly released AI model Claude Opus 4.8 as a tool during the security review, signaling the growing role of advanced AI in uncovering complex cryptographic vulnerabilities.

marsbit1 ч. назад

Behind ZEC's Over 30% Plunge: An 'Unlimited Minting' Vulnerability with No Way to Prove if It Was Ever Exploited

marsbit1 ч. назад

Торговля

Спот
Фьючерсы

Популярные статьи

Неделя обучения по популярным токенам (2): 2026 может стать годом приложений реального времени, сектор AI продолжает оставаться в тренде

2025 год — год институциональных инвесторов, в будущем он будет доминировать в приложениях реального времени.

1.8k просмотров всегоОпубликовано 2025.12.16Обновлено 2025.12.16

Неделя обучения по популярным токенам (2): 2026 может стать годом приложений реального времени, сектор AI продолжает оставаться в тренде

Обсуждения

Добро пожаловать в Сообщество HTX. Здесь вы сможете быть в курсе последних новостей о развитии платформы и получить доступ к профессиональной аналитической информации о рынке. Мнения пользователей о цене на AI (AI) представлены ниже.

活动图片

Топ вопросы

Популярные категорииright-arrow

Популярные тегиright-arrow