Radiant Capital halts Arbitrum markets after reported $4.5M flash loan attack

Cointelegraph发布于2024-01-02更新于2024-01-03

文章摘要

Cross-chain lending protocol Radiant Capital has paused its lending and borrowing markets on Arbitrum after receiving reports of a $4.5 million exploit affecting one of its newly created USDC Coin (USDC) markets.

Cross-chain lending protocol Radiant Capital has paused its lending and borrowing markets on Arbitrum after receiving reports of a $4.5 million exploit affecting one of its newly created USDC Coin (USDC) markets.
“Today, we received a report of an issue with the newly created native USDC market on Arbitrum,” said Radiant in a Jan. 3 post on X (formerly Twitter), which they added was later validated by Radiant developers and the wider cybersecurity community.
Today, we received a report of an issue with the newly created native USDC market on Arbitrum. After validation by Radiant developers and the wider Web 3 security community, the Radiant DAO Council paused lending/borrowing markets on Arbitrum temporarily while this is…
— Radiant Capital (@RDNTCapital) January 3, 2024
Blockchain security firm Beosin described the exploit as a flash loan attack — with the attacker exploiting a “rounding issue” in the codebase, “which led to a cumulative precision error.”
This ultimately allowed the “attacker to profit through repeated deposit() and withdraw() operations,” it wrote in a Jan. 3 post on X.
An earlier Jan. 2 post from PeckShield also identified the issue as caused by a “known rounding issue” in the current Compound/Aave codebase.
“The root cause is not new: It basically exploits a time window when a new market is activated in a lending market (forked from the popular Compound/Aave),” it added.
Radiant Capital @RDNTCapital was under a flash loan attack with a loss of $4.5M.
Attacker: https://t.co/L7fXlF8VXP

The attacker manipulated the index parameter (which later served as a denominator) to become extremely large. The contract has a rounding issue in its… pic.twitter.com/8AdY7pjaKE
— Beosin Alert (@BeosinAlert) January 3, 2024
The exploiter managed to siphon a total of $4.5 million in Ether (ETH) from the protocol, according to data from Arbitrum block explorer Arbiscanner.
Radiant has since paused lending and borrowing markets on Arbitrum, and reassured investors that no additional funds were currently at risk. It promised a detailed postmortem, and pledged to restore normal operations once the investigation was completed.
“As a reminder, no action can be taken until the markets are unpaused on Arbitrum,” Radiant added.
Related: Orbit Bridge hack pushes December crypto theft to nearly $100M
Meanwhile, Crypto X has already been flooded with fake Radiant Capital accounts posting phishing links purporting to help users revoke approvals.

A fake Radiant Capital account attempts to trick unsuspecting users into clicking phishing links. Source: XRadiant Capital is a decentralized borrowing and lending protocol with cross-chain functionality built using LayerZero technology. The protocol currently has around $315 million in total value locked, according to DefiLlama.
Magazine: DeFi’s billion-dollar secret: The insiders responsible for hacks

你可能也喜欢

空中云汇转向:从一年前鄙夷稳定币,到如今高调资本入局

跨境支付巨头空中云汇(Airwallex)近日领投了代币化金融结算网络Metal的种子轮融资,此举引发关注,因其创始人Jack Zhang一年前曾公开批评稳定币,质疑其降低汇款成本的效用,并认为加密货币缺乏实际用例。然而,此次投资标志着其态度发生显著转变。 Metal是一个面向代币化金融的全球结算网络与Layer-1区块链,旨在支持股票、债券、基金等各类金融资产的代币化结算,目标市场达十万亿美元级。空中云汇通过投资将为其支付网络引入代币化资产,结合自身在法币通道、全球支付场景方面的优势,形成战略协同。 尽管Jack Zhang在投资后仍坚持认为稳定币与加密货币本质不同,强调稳定币是法币的代币化形式,但其行动已表明了对稳定币及代币化赛道价值的重新评估。这一转变并非孤例,近年来传统金融巨头如摩根大通、Visa、Stripe等均在积极布局稳定币支付和代币化网络,反映出主流金融体系对链上结算效率革命的接纳。 空中云汇的投资逻辑并非全然认同加密货币,而是基于战略考量。面对稳定币在新兴市场、企业金融和链上结算等领域带来的结构性机会,以及可能重塑支付行业竞争格局的趋势,提前入场布局成为必要选择。这回答了其一年前的质疑:稳定币的价值至少已值得资本下注,而非置身事外。

marsbit38分钟前

空中云汇转向:从一年前鄙夷稳定币,到如今高调资本入局

marsbit38分钟前

第一个用物理做计算原语的大规模生成模型Un-0来了,或将AI能耗降低1000倍?

在AI能耗问题日益成为行业瓶颈的背景下,前Databricks AI负责人Naveen Rao创立的Unconventional AI公司发布了其首个大规模生成模型Un-0。该模型的核心创新在于利用“模拟耦合振子系统”作为计算原语,通过物理系统的自然演化过程来完成图像生成任务,旨在将AI推理能耗降低至现有系统的千分之一。 Un-0的工作原理基于Kuramoto振子模型。成千上万个振子根据其固有频率和相互间的耦合关系,在时间维度上自然演化、同步或形成特定模式。模型的训练主要学习振子间的耦合矩阵和固有频率。生成图像时,系统首先随机初始化振子相位,然后通过条件振子输入类别标签(如“火山”)进行引导,接着让物理系统自由演化,最后在特定时刻读取振子相位状态,并通过一个轻量级解码器将其转换为图像像素。 在ImageNet 64×64数据集上,拥有3.22亿参数的Un-0模型取得了FID 6.74的成绩,其生成质量已接近一些早期主流图像生成模型(如BigGAN、iDDPM)发布时的水平,但与当前最先进的传统模型(如EDM)相比仍有差距。Unconventional AI强调,Un-0并非旨在立即成为性能冠军,而是一个“概念验证”,证明了利用物理动力学系统执行现代AI大规模生成任务是可行的。 该技术路线的长远目标是构建一种新型的非传统计算硬件,将计算与记忆合并于同一物理实体中,从而避免冯·诺依曼架构中数据搬运带来的巨大能耗。Un-0的出现,标志着AI计算范式向利用物理系统固有动力学特性转变迈出了重要一步,为未来实现能效数量级提升的AI硬件开辟了新路径。

marsbit57分钟前

第一个用物理做计算原语的大规模生成模型Un-0来了,或将AI能耗降低1000倍?

marsbit57分钟前

辣评|「最有情绪价值」的帖子!STRC 会是下一个 LUNA 吗?

**《辣评》本周聚焦三件事:散户情绪、STRC脱锚风险与交易员川沐的“矛盾人设”。** **1. 本月“最有情绪价值”的帖子** 6月加密货币市场大跌,BTC跌破6万美元。Reddit上一条“去你的恐惧、不确定和怀疑!”的帖子引发散户共鸣,评论区成为情绪宣泄地,有人喊“钻石手加仓”,也有人调侃“不卖都是浮亏”,展现了熊市中的集体心理。 **2. STRC脱锚,会是下一个LUNA吗?** Strategy公司发行的永续优先股STRC价格跌至76.2美元,较面值脱锚约25%。投资者担忧其每年12亿美元的股息支付能力——Strategy当前美元储备约14亿美元,仅够支付一年。关键区别在于,CEO Saylor并无支付股息的强制性义务,因此STRC虽可能下跌,但不会像LUNA那样归零。不过,若市场对支付能力失去信心,其普通股MSTR也可能受牵连。Strategy CEO已购入100万美元STRC以示信心,但类似LUNA崩盘前的言论也引发了市场警觉。 **3. 川沐的1828条推文与“矛盾人设”** 交易员川沐因在ORDI行情和2025年做空山寨币中获利百万而闻名。分析其推文发现,其核心方法是追问系统瓶颈(如AI浪潮中的存储、光通信等环节),并据此迁移投资主线。然而,他也被质疑曾“喊单”部分币种后出现“收割”行情,且中途卖出股票未告知跟风者,呈现“盈利展示”与私下操作不一的矛盾形象,提醒投资者保持独立判断。 **总结**:本周市场情绪低迷,但社区互勉;STRC面临股息支付考验,虽与LUNA本质不同,风险犹存;川沐的案例则揭示了光鲜战绩背后的复杂性与跟单风险。

Foresight News1小时前

辣评|「最有情绪价值」的帖子!STRC 会是下一个 LUNA 吗?

Foresight News1小时前

交易

现货
活动图片