Australian Treasury to Question Regulator Over HyperVerse Crypto Scheme: Report

CoinDeskPolicy發佈於 2024-01-04更新於 2024-01-05

文章摘要

"It seemed pretty clear that there should have been concerns raised about… this operation,” Stephen Jones said.

Australia's Assistant Treasurer and Minister for Financial Services Stephen Jones has said he would be asking the Australian Securities and Investments Commission (ASIC) why it didn't warn consumers about the HyperVerse crypto scheme like other nations did, according to the Guardian.

The United Kingdom, New Zealand, Canada, Germany and Hungary, among others, issued warnings about the scheme as early as 2021, the report said.

“This type of scheme works by convincing innocent people to invest their money into a product that might not exist, with the only source of income being money from new investors,” Jones reportedly said. "I simply don’t know why a warning wasn’t issued. It seemed pretty clear that there should have been concerns raised about… this operation.”

Advertisement
Advertisement

The HyperVerse crypto scheme resulted in thousands of people losing millions of dollars, according to an investigation by Guardian Australia last month. The scheme was run by an entity called HyperTech and was promoted and run by CEO Steven Reece Lewis who does not appear to exist.

The founders of HyperTech, Australian entrepreneur Sam Lee and his business partner Ryan Xu, also founded the collapsed Australian bitcoin company Blockchain Global which owes creditors $58m. The liquidators alerted ASIC about Lee and Xu for breaking the law but the regulator said it does not intend to take action at this time, the Guardian reported.

ASIC did not immediately respond to a CoinDesk request for comment. HyperTech could not be reached for comment.

Edited by Oliver Knight.

你可能也喜歡

拖更三年,北大校友翁荔最新长文刷屏

前OpenAI副总裁翁荔(Lilian Weng)发表了一篇关于AI扩展定律(Scaling Laws)的深度分析文章。文章指出,这条指导了AI行业数百亿美元投入的核心定律,远比人们想象的更为脆弱。 文章回顾了Scaling Laws的基本思想,即模型性能随规模扩大而可预测地提升。然而,OpenAI与DeepMind在关键问题上得出了相反结论:给定算力,资源应更多分配给模型还是数据?OpenAI的Kaplan团队认为模型增长应更快,而DeepMind的Chinchilla团队则认为应等比增长。后来研究发现,这一分歧源于参数统计口径的差异和实验规模不足,导致Kaplan的结论仅适用于小规模场景。 更关键的是,被行业广泛采纳的Chinchilla最优配比公式本身也存在方法论瑕疵。2024年有团队复现发现,其损失函数因取均值而非求和,导致优化器提前停止,输出并非最优解。此外,用于外推的关键参数精度不足,放大了误差。 文章进一步指出,经典Scaling Laws的根本前提——高质量数据无限供应——正在崩塌。数据重复训练不可避免,新研究引入了惩罚项来修正公式,并发现大模型对数据重复更敏感。这解释了行业为何转向强化学习、测试时计算和合成数据等新路径。 翁荔的博客以其清晰深入的技术解析著称,这篇文章历时三年完成。她于2025年联合创立了新公司Thinking Machines Lab。文章强调,下一代AI的进步不仅依赖算力规模,更取决于对这些基础定律细节更精确的理解与运用。

marsbit1 小時前

拖更三年,北大校友翁荔最新长文刷屏

marsbit1 小時前

交易

現貨
合約
活动图片