Polymarket Denies Alleged Data Breach Amid Dark Web Leak Claims

TheNewsCrypto发布于2026-04-29更新于2026-04-29

There have been rumors circulating that a hacker on the dark web leaked what they said was a treasure mine of confidential user information from Prediction markets platform Polymarket, but the company has disputed the claims.

On Tuesday, numerous X accounts that monitor the dark web—including cybersecurity firm Vecert Analyzer—shared pictures from DarkForums showing a hacker posing as “xorcat” and claiming to have compromised Polymarket.

Data Already Accessible Onchain

The hacker group xorcat said in their post that they had obtained over 300,000 data, 10,000 of which were user profiles complete with usernames, photos, base addresses, and proxy wallets. Polymarket condemned the data breach allegations as “complete and utter nonsense” and said that the hacker’s provided information is already accessible online.

Many in the cryptocurrency field were on high alert in April after a dramatic increase in breaches and attacks using cryptocurrency. Web3 projects lost $482 million in 44 incidents during the first quarter of 2026, according to blockchain security start-up Hacken. The study was published earlier this month.

One advantage of being on chain is that all of our data is publicly auditable. This is a feature, not a bug, according to the prediction market. There was no data breach; all of its publicly available endpoints and on-chain data are accessible. One may obtain the data for free using its APIs instead of paying for it, the prediction market platform stated.

The purported hacker said that Polymarket lacked a bug bounty program, which led to the disclosure of sensitive information. On the other hand, as of Wednesday, 446 reports have been received by Polymarket’s live bug bounty program, which began on April 16.

According to Xorcat, data was extracted from Polymarket’s Gamma and CLOB APIs using pagination bypass, CORS misconfiguration, and undocumented API endpoints. This hacker went on the record as having broken into additional prediction markets and was planning to leak the data in the coming days.

Highlighted Crypto News Today:

Skynet Report Flags Shift in Crypto Regulation Risk Toward AML Enforcement

TagsAltcoinBlockchain

你可能也喜欢

英特尔 CEO 陈立武首次播客访谈:我们的目标是“5-10 年 10 倍”,押注先进封装、玻璃基板和人工钻石

英特尔CEO陈立武在接受播客访谈时,提出了“5-10年实现10倍回报”的股东回报目标,并系统阐述了其领导下的转型战略。面对传统工艺微缩接近物理极限的挑战,他将突破点聚焦于先进封装技术(如EMIB)、新型基板材料(如玻璃基板)以及氮化镓、碳化硅和人工合成钻石等半导体新材料,旨在通过材料科学与封装创新延续性能增长。 陈立武将转型过程分为“爬、走、跑”三个阶段。目前英特尔处于夯实基础的“爬行”阶段,首要任务是稳固资产负债表、聚焦简化产品线、并倾听客户需求。他特别指出,智能体AI和推理场景的爆发正带动CPU需求强劲回升。 在代工业务方面,他强调这是一门“信任的生意”,核心是提升良率、缺陷密度和周期时间等基础能力,以服务客户并保障美国本土供应链安全。他同时透露,与埃隆·马斯克合作的Terafab项目进展顺利,双方正共同应对AI算力增长带来的半导体基础设施挑战。 陈立武认为,市场目前低估了英特尔的长期潜力。他预计到2030-2032年,外界将真正看到其在PC基本盘之外,于边缘计算、物理AI与智能体AI等新兴市场的价值。通过整合XPU(混合架构)、先进封装与代工能力,为不同工作负载提供定制化芯片解决方案,是其为英特尔锚定的长期战略方向。

marsbit1小时前

英特尔 CEO 陈立武首次播客访谈:我们的目标是“5-10 年 10 倍”,押注先进封装、玻璃基板和人工钻石

marsbit1小时前

AI 成绩单背后,藏着一位华人“出题人”

AI领域重要的基准评测MMLU-Pro、MMMU、MMMU-Pro背后,都站着同一位“出题人”——加拿大滑铁卢大学助理教授陈文虎。 随着大语言模型能力快速提升,旧的评测基准如MMLU逐渐“失灵”,顶尖模型得分趋近满分,难以区分高下。为此,陈文虎团队于2024年推出MMLU-Pro。它通过将选项扩至10个、增加推理题比例、剔除简单题等方式,使模型准确率相较旧基准显著下降16%-33%,成绩波动更小,有效拉开了模型间的真实差距。 在多模态评测方面,陈文虎团队推出的MMMU基准包含1.15万道需结合图像与专业知识的复杂题目,即便是当时最强的GPT-4V和Gemini Ultra准确率也未超过60%。后续的MMMU-Pro则进一步堵住模型仅凭文本猜答案的漏洞,强制其进行真正的多模态理解。 陈文虎的研究方向长期聚焦于复杂信息理解与推理。他曾于谷歌DeepMind参与Gemini多模态模型的研发与评估工作,这让他深谙模型能力增长路径与评估盲区。他创立的“老虎实验室”不仅做评测,也研发视频理解与生成等模型,这种“既做题也出题”的经历,使其能更精准地设计出触及模型能力边界的评估体系。 如今,陈文虎已加入Meta超级智能实验室,继续专注于多模态预训练与评估工作。他的故事反映了在AI浪潮中,众多华人研究者正深度参与并塑造着行业发展的核心基础工作。

marsbit3小时前

AI 成绩单背后,藏着一位华人“出题人”

marsbit3小时前

交易

现货
合约
活动图片