Tech Share Share this article Copy linkX (Twitter)LinkedInFacebookEmail Anthropic Research Shows AI Agents Ar Tech Share Share this article Copy linkX (Twitter)LinkedInFacebookEmail Anthropic Research Shows AI Agents Ar

Anthropic Research Shows AI Agents Are Closing In on Real DeFi Attack Capability

Share
Share this article
Copy linkX (Twitter)LinkedInFacebookEmail

Anthropic Research Shows AI Agents Are Closing In on Real DeFi Attack Capability

Models tested by MATS and the Anthropic Fellows program generated turnkey exploit scripts and identified fresh vulnerabilities, suggesting automated exploitation is becoming technically and economically viable.

By Sam Reynolds|Edited by Sheldon Reback
Dec 2, 2025, 9:11 a.m.
AI agents are now capable of identifying flaws in smart contracts. (Markus Winkler/Unsplash modified by CoinDesk)

What to know:

  • AI agents are now capable of finding and exploiting vulnerabilities in smart contracts, posing a potential threat as demonstrated by recent research.
  • Models like GPT-5 and Sonnet 4.5 have successfully simulated exploits, revealing the feasibility of autonomous attacks in decentralized finance (DeFi).
  • Researchers warn that as AI models become cheaper and more advanced, the risk of automated exploitation extends beyond DeFi to broader software and infrastructure vulnerabilities.

AI agents are getting good enough at finding attack vectors in smart contracts that they can already be weaponized by bad actors, according to new research published by the Anthropic Fellows program.

A study by the ML Alignment & Theory Scholars Program (MATS) and the Anthropic Fellows program tested frontier models against SCONE-bench, a dataset of 405 exploited contracts. GPT-5, Claude Opus 4.5 and Sonnet 4.5 collectively produced $4.6 million in simulated exploits on contracts hacked after their knowledge cutoffs, offering a lower bound on what this generation of AI could have stolen in the wild.

STORY CONTINUES BELOW
Don't miss another story.Subscribe to the The Protocol Newsletter today. See all newsletters
Sign me up
(Anthropic Labs & MATS)

The team found that frontier models did not just identify bugs. They were able to synthesize full exploit scripts, sequence transactions and drain simulated liquidity in ways that closely mirror real attacks on the Ethereum and BNB Chain blockchains.

The paper also tested whether current models could find vulnerabilities that had not yet been exploited.

GPT-5 and Sonnet 4.5 scanned 2,849 recently deployed BNB Chain contracts that showed no signs of prior compromise. Both models uncovered two zero-day flaws worth $3,694 in simulated profit. One stemmed from a missing view modifier in a public function that allowed the agent to inflate its token balance.

Another allowed a caller to redirect fee withdrawals by supplying an arbitrary beneficiary address. In both cases, the agents generated executable scripts that converted the flaw into profit.

Although the dollar amounts were small, the discovery matters because it shows that profitable autonomous exploitation is technically feasible.

The cost to run the agent on the entire set of contracts was only $3,476, and the average cost per run was $1.22. As models become cheaper and more capable, the economics tilt further toward automation.

Researchers argue that this trend will shorten the window between contract deployment and attack, especially in DeFi environments where capital is publicly visible and exploitable bugs can be monetized instantly.

While the findings focus on DeFi, the authors warn that the underlying capabilities are not domain-specific.

The same reasoning steps that let an agent inflate a token balance or redirect fees can apply to conventional software, closed-source codebases, and infrastructure that supports crypto markets.

As model costs fall and tool use improves, automated scanning is likely to expand beyond public smart contracts to any service along the path to valuable assets.

The authors frame the work as a warning rather than a forecast. AI models can now perform tasks that historically required highly skilled human attackers, and the research suggests that autonomous exploitation in DeFi is no longer hypothetical.

The question now for crypto builders is how quickly defense can catch up.

Artificial IntelligenceHackDeFiDecentralized Finance

More For You

Protocol Research: GoPlus Security

Commissioned byGoPlus

What to know:

  • As of October 2025, GoPlus has generated $4.7M in total revenue across its product lines. The GoPlus App is the primary revenue driver, contributing $2.5M (approx. 53%), followed by the SafeToken Protocol at $1.7M.
  • GoPlus Intelligence's Token Security API averaged 717 million monthly calls year-to-date in 2025 , with a peak of nearly 1 billion calls in February 2025. Total blockchain-level requests, including transaction simulations, averaged an additional 350 million per month.
  • Since its January 2025 launch , the $GPS token has registered over $5B in total spot volume and $10B in derivatives volume in 2025. Monthly spot volume peaked in March 2025 at over $1.1B , while derivatives volume peaked the same month at over $4B.
View Full Report

More For You

Ethereum Devs Push ZK ‘Secret Santa’ System Toward Deployment

The proposed protocol uses zero-knowledge proofs to verify sender–receiver relationships without revealing identities.

What to know:

  • Ethereum developers are working on a zero-knowledge protocol to enhance privacy in on-chain interactions, starting with a Secret Santa-style matching system.
  • The protocol uses zero-knowledge proofs and transaction relayers to maintain anonymity and prevent Sybil attacks in blockchain transactions.
  • This initiative is part of a larger effort to create privacy frameworks for Ethereum, applicable in areas like anonymous voting and private token distributions.
Read full story
Latest Crypto News

Unlimit Debuts Stable.com, a Decentralized Clearing House Built for Stablecoins

XRP, Bitcoin on the Edge; Will Santa Abandon Nasdaq?

Ethereum Devs Push ZK ‘Secret Santa’ System Toward Deployment

Sanctioned Cambodian Lender Huione, Linked to Illicit Crypto, Halts Business After Bank Run: Report

Dogecoin Wicks Below Key Support — Fakeout or Start of Larger Correction?

Bitcoin May Dump to $65K or Below, Spelling Trouble for ETH, XRP, ADA and Other Majors

Top Stories

Bitcoin May Dump to $65K or Below, Spelling Trouble for ETH, XRP, ADA and Other Majors

Ethereum Devs Push ZK ‘Secret Santa’ System Toward Deployment

Attention Bitcoin Bulls: The U.S. 10-Year Yield Isn't Budging Despite Fed Rate Cut Hopes

Vanguard Opens Platform to Crypto ETFs in Major Shift: Bloomberg

U.S. House Lawmakers Detail Grievances Over Government's 'Choke Point 2.0'

Bitcoin Traders Bet on Sub-$80K New Year: Derive

Market Opportunity
null Logo
null Price(null)
--
----
USD
null (null) Live Price Chart
Disclaimer: The articles reposted on this site are sourced from public platforms and are provided for informational purposes only. They do not necessarily reflect the views of MEXC. All rights remain with the original authors. If you believe any content infringes on third-party rights, please contact service@support.mexc.com for removal. MEXC makes no guarantees regarding the accuracy, completeness, or timeliness of the content and is not responsible for any actions taken based on the information provided. The content does not constitute financial, legal, or other professional advice, nor should it be considered a recommendation or endorsement by MEXC.

You May Also Like

A Radical Neural Network Approach to Modeling Shock Dynamics

A Radical Neural Network Approach to Modeling Shock Dynamics

This paper introduces a non-diffusive neural network (NDNN) method for solving hyperbolic conservation laws, designed to overcome the shortcomings of standard Physics-Informed Neural Networks (PINNs) in modeling shock waves. The NDNN framework decomposes the solution domain into smooth subdomains separated by discontinuity lines, identified via Rankine-Hugoniot conditions. This approach enables accurate tracking of entropic shocks, shock generation, and wave interactions, while reducing the diffusive errors typical in PINNs. Numerical experiments validate the algorithm’s potential, highlighting its promise for extending shock-wave computations to higher-dimensional problems.
Share
Hackernoon2025/09/19 18:38
A Netflix ‘KPop Demon Hunters’ Short Film Has Been Rated For Release

A Netflix ‘KPop Demon Hunters’ Short Film Has Been Rated For Release

The post A Netflix ‘KPop Demon Hunters’ Short Film Has Been Rated For Release appeared on BitcoinEthereumNews.com. KPop Demon Hunters Netflix Everyone has wondered what may be the next step for KPop Demon Hunters as an IP, given its record-breaking success on Netflix. Now, the answer may be something exactly no one predicted. According to a new filing with the MPA, something called Debut: A KPop Demon Hunters Story has been rated PG by the ratings body. It’s listed alongside some other films, and this is obviously something that has not been publicly announced. A short film could be well, very short, a few minutes, and likely no more than ten. Even that might be pushing it. Using say, Pixar shorts as a reference, most are between 4 and 8 minutes. The original movie is an hour and 36 minutes. The “Debut” in the title indicates some sort of flashback, perhaps to when HUNTR/X first arrived on the scene before they blew up. Previously, director Maggie Kang has commented about how there were more backstory components that were supposed to be in the film that were cut, but hinted those could be explored in a sequel. But perhaps some may be put into a short here. I very much doubt those scenes were fully produced and simply cut, but perhaps they were finished up for this short film here. When would Debut: KPop Demon Hunters theoretically arrive? I’m not sure the other films on the list are much help. Dead of Winter is out in less than two weeks. Mother Mary does not have a release date. Ne Zha 2 came out earlier this year. I’ve only seen news stories saying The Perfect Gamble was supposed to come out in Q1 2025, but I’ve seen no evidence that it actually has. KPop Demon Hunters Netflix It could be sooner rather than later as Netflix looks to capitalize…
Share
BitcoinEthereumNews2025/09/18 02:23
Headwind Helps Best Wallet Token

Headwind Helps Best Wallet Token

The post Headwind Helps Best Wallet Token appeared on BitcoinEthereumNews.com. Google has announced the launch of a new open-source protocol called Agent Payments Protocol (AP2) in partnership with Coinbase, the Ethereum Foundation, and 60 other organizations. This allows AI agents to make payments on behalf of users using various methods such as real-time bank transfers, credit and debit cards, and, most importantly, stablecoins. Let’s explore in detail what this could mean for the broader cryptocurrency markets, and also highlight a presale crypto (Best Wallet Token) that could explode as a result of this development. Google’s Push for Stablecoins Agent Payments Protocol (AP2) uses digital contracts known as ‘Intent Mandates’ and ‘Verifiable Credentials’ to ensure that AI agents undertake only those payments authorized by the user. Mandates, by the way, are cryptographically signed, tamper-proof digital contracts that act as verifiable proof of a user’s instruction. For example, let’s say you instruct an AI agent to never spend more than $200 in a single transaction. This instruction is written into an Intent Mandate, which serves as a digital contract. Now, whenever the AI agent tries to make a payment, it must present this mandate as proof of authorization, which will then be verified via the AP2 protocol. Alongside this, Google has also launched the A2A x402 extension to accelerate support for the Web3 ecosystem. This production-ready solution enables agent-based crypto payments and will help reshape the growth of cryptocurrency integration within the AP2 protocol. Google’s inclusion of stablecoins in AP2 is a massive vote of confidence in dollar-pegged cryptocurrencies and a huge step toward making them a mainstream payment option. This widens stablecoin usage beyond trading and speculation, positioning them at the center of the consumption economy. The recent enactment of the GENIUS Act in the U.S. gives stablecoins more structure and legal support. Imagine paying for things like data crawls, per-task…
Share
BitcoinEthereumNews2025/09/18 01:27