Lido adjusts EarnETH's fee structure, lowering the management fee to 0.2% and increasing the proportion of performance fees.

X has sued Vivek Kumar Sen, Zamyang Sherpa, and other unnamed individuals in the UK High Court, accusing them of operating a coordinated bot network that manipulated engagement data to defraud at least ￡207,400 (approximately $278,000) from X’s Creator Revenue Sharing Program. X alleges the defendants operated 6 primary accounts and at least 3 secondary accounts, regularly posting highly similar Bitcoin content and generating fake engagement via mutual likes, retweets, and other interactions between the accounts. Stripe accounts linked to some of the accounts were found to use names and bank accounts mismatched with the actual operators to obscure fund sources. X states the network participated in the Creator Revenue Sharing Program from 2023 to 2026; the program, which allocated ad revenue based on engagement metrics including likes, replies, and retweets, was officially shut down on September 7 and replaced by Original Content Rewards, which explicitly excludes "artificially generated" fake engagement. X suspended the 9 accounts in question on August 18, demanding the defendants return at least $278,000, plus claims for investigation costs, damages, interest, and legal fees. As of September 21, no defendants have submitted a defense in the case.

Bitcoin surged past the key resistance level of $82,000 to hit a high of $87,300, triggering liquidations of around $750 million in bearish crypto derivative positions, with buying pressure from short covering further fueling the price rally. Since the breakout, Bitcoin’s derivative open interest has added roughly $2 billion, indicating a rapid return of market leverage. Nansen analyst Nicolai Sondergaard noted that “prices are turning bullish faster than position adjustments.” U.S. spot Bitcoin ETFs, which had faced headwinds from the CLARITY Act and Federal Reserve rate hikes, saw a combined outflow of $746 million on Tuesday and Wednesday, followed by inflows of $160 million and $433 million on Thursday and Friday respectively. The average cost basis for U.S. Bitcoin ETF holders is around $82,200; with Bitcoin breaking above this level, ETF investors as a whole have returned to profitability. Looking ahead, Sondergaard identifies $87,000 as the next key level, with $90,000 and $92,000 as subsequent targets; Wintermute trader Jasper De Maere also believes Bitcoin could test $90,000. However, he pointed out that breaking above the 50-week moving average is merely a positive signal, and significant volatility could still occur in the short term. The market is now closely watching whether spot and ETF inflows can sustain the rally. Sondergaard warned that if spot demand fails to keep pace with the growth in derivative leverage, the rally could become leverage-driven and reverse rapidly amid rising U.S. Treasury yields or geopolitical shocks. Wintermute advised focusing on ETF flows, perpetual contract open interest and funding rates in the coming days, as well as the options expiration this Friday.

The U.S. Department of Justice announced on Sept. 21 that Iris Rabaya Au, a 37-year-old woman from Irvine, Orange County, California, was sentenced to 18 months in federal prison for failing to report over $2.6 million in illegal income to the Internal Revenue Service (IRS). She was ordered to pay $1.484 million in restitution, and authorities seized multiple luxury and high-performance vehicles, designer handbags, and three "Godfather" sculptures, among other assets. Au is the ex-girlfriend of crypto fraudster Adam Iza. Prosecutors alleged that between 2020 and 2024, Iza fraudulently obtained Meta ad accounts and lines of credit, then sold access to those accounts to generate millions in unreported income. Acting on Iza’s instructions, Au set up shell companies and bank accounts, using the funds to pay roughly $1 million to Los Angeles County Sheriff’s Department (LASD) deputies, and to purchase or rent luxury homes, vehicles, jewelry, and clothing. The money also covered nearly $10 million in entertainment expenses for the pair, and Iza used part of the proceeds to buy approximately $16 million in cryptocurrency. Between 2020 and 2023, Au transferred over $2.6 million to personal bank accounts but did not accurately report the funds on her federal tax returns. She pleaded guilty in March 2025 to one count of filing false tax returns. Iza pleaded guilty in January 2025 to charges including conspiracy to violate rights, wire fraud, and tax evasion; his sentencing hearing is set for Oct. 5. He was also sentenced to 15 years in federal prison in August 2024 for his involvement in an attempted bitcoin robbery and kidnapping case in Connecticut.

Robinhood CEO Vlad Tenev said crypto-related prediction contracts are taking an increasing share of Robinhood’s prediction market, and he expects sports contracts could fall to a minority in the coming years. In an interview with CNBC’s *Mad Money*, Tenev noted that sports contracts were once a key entry point for driving growth in the prediction market business, but other categories like crypto are gaining "disproportionate share". Robinhood’s event contract revenue in Q2 2026 surged more than 10-fold year-over-year to $156 million, with contract trading volume hitting 4.7 billion in August—roughly 15 times the level in the same period of 2025. Robinhood currently builds its prediction market on Kalshi, offers related services via Rothera, a joint venture formed with Susquehanna, and has also taken an equity stake in Crypto.com and its prediction market subsidiary OG.com. Tenev added that crypto prediction contracts allow users to directly translate their judgments on events like markets and policies into trades.

Beating AI News Flash, according to FT reports, OpenAI on Monday called on the U.S. government to lead the establishment of international AI standards, and suggested that countries coordinate to assess the capabilities and risks of AI models. OpenAI CEO Sam Altman will present the relevant proposal to the UN Security Council this week. OpenAI proposed leveraging the existing network of national AI safety research institutions to build a unified international AI model evaluation standard, and called on the U.S. to set up secure communication channels with overseas countries to share emerging threats in the AI field. OpenAI specifically warned of potential new risks from Recursive Self-Improvement (RSI), meaning an increasing number of AI model training tasks are completed by other AI systems. Without proper safety measures, AI could iterate rapidly to a level that humans cannot effectively supervise, gradually causing humanity to lose actual control over AI development. The proposal comes amid growing concerns about AI safety. Both Anthropic and OpenAI have recently disclosed incidents of AI models breaching external systems. Earlier this month, OpenAI CEO Sam Altman, Anthropic CEO Dario Amodei, and SpaceX CEO Elon Musk also called for slowing down AI development. U.S. President Donald Trump, however, opposes slowing AI progress, stating that "whoever wins the AI race will win everything".

Beating AI Express News: xAI officially launched Grok 4.7 on Monday, with the company noting it delivers "significant improvements" over Grok 4.6 at the same price and speed points. Grok 4.7 had been delayed at least five times prior to launch, and is now available on the Grok app, Cursor, Grok Build, and the xAI API. The model has a parameter size of 2.1 trillion, a 40% jump from Grok 4.6’s 1.5 trillion, with input and output pricing set at $2 and $6 per million tokens respectively. xAI also incorporated supplementary training data from SpaceX, including Starlink satellite telemetry, manufacturing records, and engineering failure logs, to boost the model’s reasoning capabilities for hardware and physical systems. In benchmarks, Grok 4.7 ranked second in both GDPval and AA-Briefcase, scoring 1695 and 1657 respectively, trailing Claude Fable 5.1’s 1735 and 1678. It also placed second in the electrical engineering benchmark EEBench, behind only GPT-6 Astra. CursorBench 4.0 data shows Grok 4.7 has higher task costs than GPT-6 Astra and Claude Sonnet 5, while its overall performance still lags behind Fable 5.1. Musk previously stated Grok 4.7 is roughly on par with Claude Opus 5.0, though it has not reached the level of the newer Opus 5.1, and revealed Grok 4.8, 4.9, and Grok 5 are in active development.