Moonwell: Recovery Plan Submitted to Governance Forum and Partial Compensation Initiated

U.S. President Donald Trump posted two AI-generated videos on Truth Social late Sunday, with the caption "Khark Island was blown to smithereens". The videos show oil fields, storage tanks and docks being hit by missile strikes and exploding. However, there is currently no evidence that the U.S. military actually attacked Khark Island. U.S. Central Command stated that the U.S. military carried out "limited and precise" strikes on an Iranian Revolutionary Guard Corps (IRGC) mine-laying unit on Sunday, targeting the area near Larak Island in the Strait of Hormuz, aimed at eliminating an imminent threat to commercial shipping and civilian vessels. Notably, Khark Island and Larak Island are far apart; the former, located deep in the Persian Gulf, is Iran’s main oil export hub. It remains unclear whether the AI videos posted by Trump were produced by his team, or whether he confused Larak Island with Khark Island. The IRGC subsequently claimed it had launched ballistic missiles at two U.S. military bases in Jordan. Jordan’s authorities said its air defense systems intercepted 8 missiles, with no casualties reported. Hamid Bovard, CEO of the National Iranian Oil Company, said Monday that oil operations at Iran’s Khark Island have not stopped, denying Trump’s earlier post claiming the island’s key oil hub was destroyed.

Blockchain security firm Blockaid said it has detected an exploit targeting More Markets (operated by More Labs) on the Flow EVM. Attackers exploited Ankr’s bonded LST (liquid staking token) and E-Mode mechanism to successfully drain More Markets’ WFLOW lending reserves. According to Blockaid’s monitoring, the attackers transferred approximately 15.5 million WFLOW tokens from the mFlowWFLOW reserve, with the detected impact amounting to around $9.3 million. The relevant attack transaction cluster also includes post-exploit fund transfers. The exact losses from the incident and the attackers’ fund destinations are still under further confirmation.

Wall Street's so-called "Fed Whisperer" Nick Timiraos said that Fed Chair Walsh delivered clear hawkish signals at last week's annual global central bank conference, noting that current financial conditions are hardly restrictive, and recent improvements in inflation are insufficient to prove underlying price pressures have materially eased. As a result, markets have lifted their expectations for a September rate hike by the Fed, with some institutions even forecasting one rate hike each in September and December this year. Timiraos pointed out that the August CPI data, set to be released on September 11, will be a key determinant of the Fed's September policy path. If inflation continues to cool, the Fed may have reason to hold interest rates steady; however, if CPI rebounds sharply, Walsh will face greater policy pressure, and keeping rates unchanged would be hard to justify. Meanwhile, the hawk-dove divide within the Fed has intensified. Three officials dissented at the last FOMC meeting, while other officials openly supported rate hikes. Markets are currently closely monitoring inflation data, as well as whether tariffs, Middle East tensions and the AI investment boom will further fuel price pressures.

Official announcement from Pons: Its cumulative trading volume has surpassed $4 billion. The platform’s native token, PONS, has reached a market capitalization of $331 million, with a 24-hour trading volume exceeding $66 million. Pons has now emerged as one of the most active trading and launch platforms in the Robinhood Chain ecosystem, with market interest continuing to grow.

Amid sharp swings in AI-related stocks and intensifying short covering, Wall Street’s long-popular "buy strong, sell weak" momentum strategy is facing a sharp reversal. Data shows that since July 1, the S&P 500 Momentum Index has fallen more than 9%, while the S&P 500 Index has risen about 2.8% over the same period, pushing the momentum index toward its worst quarterly performance in 25 years. In the second quarter of this year, the S&P 500 Momentum Index surged 44%, marking its best quarterly performance on record, with a cumulative gain of 133% over the past five years, and was once one of the most crowded trading strategies among institutional investors. But entering the second half of the year, some stocks that had been heavily shorted suddenly rebounded, triggering large-scale liquidation by quantitative funds and short covering. Data from Bank of America shows that momentum trading in July this year posted the second-worst performance in nearly 40 years, second only to the 2009 financial crisis period. Goldman Sachs data shows that August 19 was one of the worst trading days for systematic long-short funds in nearly two years, with about half of losses stemming from the failure of momentum strategies. Meanwhile, net short positions held by Nasdaq 100 futures speculators have risen to near 20-year highs. Market analysts warn that while the broader U.S. stock market is still rising, AI overvaluation, surging capital expenditures by tech giants, and crowded momentum trades are exposing the market to a higher risk of sudden reversals.

Beating AI News Brief (Dongcha): AI inference platform fal engineer Rehan Sheikh has integrated H3 Max directly with Twitch. Since the model generates videos faster than viewers can stream them, the live broadcast can run while generating subsequent frames in the background, enabling theoretically endless streams. Sheikh dubbed this "Infinite Cross-Dimensional TV". MiniMax’s official account reposted the experiment, and fal later launched the H3 Max Live experiment. The live content is not pre-generated: viewers can input !prompt directly in Twitch chat. fal stated new requests appear on screen in seconds, with every scene shaped by chat input. There is no fixed script or runtime; as long as generation continues, the show runs indefinitely. The biggest hurdle for this use case was generation speed: if a 15-second video required minutes of wait time for the next segment, true live streaming was impossible. H3 Max has cut that wait time to under playback time, letting the AI prepare the next segment while viewers watch the current one. This marks the first time a video model has shifted from "waiting for output" to delivering continuous real-time content. H3 Max is a version trained by fal Research based on MiniMax’s open-source H3, optimized for fal’s inference system. Official metrics show it generates a 5-second video in under 3 seconds, with throughput ~35 times that of MiniMax’s official H3 API. Currently, in Artificial Analysis’s audio-inclusive leaderboard, its image-to-video model ranks first with 1202 Elo, while its text-to-video model ranks third with 1235 Elo, trailing only Wan 3.0 and Gemini Omni Flash.