News

On April 19, Monad co-founder Keone Hon stated, "I feel that 'liquidity pool lending protocols' should set rate limits on the supply of assets deposited as collateral. For example, if the current supply is 100 million and the supply cap is 300 million, then in the next 10 minutes, the maximum allowed increase should be to 110 million, rather than allowing a single deposit of the entire 200 million. In reality, no one needs to make such a large deposit all at once. This is crucial because when certain 'exotic assets' are attacked, the impact depends on the 'exit channel size' of that asset. Especially in many cases where attacks fall under 'infinite minting vulnerabilities', the scale of the exit that can be made essentially determines the upper limit of the attack loss. Lending protocols often serve as the largest exit channel. If a 'smart cap' were introduced, where the initial cap is slightly above the current supply and gradually adjusted to the true cap over several hours, it would have a significant effect. If such a mechanism were in place, today’s rsETH depositors could have avoided approximately $200 million in losses. This also raises the point that asset issuers themselves should support such mechanisms. If you are issuing redeemable tokens with redemption delays, you are not worried about hackers redeeming directly from you, but you need to compress the external exit path size as much as possible without affecting normal user usage. Therefore, a high supply cap should be viewed as a risk rather than a symbol of strength. For instance, the Hyperbridge DOT attack did not result in a $100 million loss because there were few exit paths; the Resolv attack resulted in a loss of $24 million instead of $200 million because the exit path size limited the loss cap. This is an obvious truth, but there are still immediately actionable measures: audit the supply caps of all assets and lower the caps when unnecessary.

Golden Ten Data reported on April 19 that Harvest Crude Oil LOF has issued a premium risk warning announcement. This fund is a QDII-LOF type listed open-end fund, mainly investing in public funds (including ETFs) that track crude oil price indices or crude oil prices as their investment objectives, and its risk level is relatively high. Recently, the secondary market trading price has continued to be higher than the fund's net asset value per unit, resulting in a significant premium. The fund manager reminds investors to pay close attention to premium risks, make prudent decisions, and avoid facing substantial investment losses caused by secondary market price corrections due to blindly purchasing at a high premium significantly deviating from the actual value of the assets. If by April 20, 2026, the premium margin has not effectively fallen back, the fund has the right to apply to the Shenzhen Stock Exchange for an intra-day temporary suspension or to extend the suspension period to warn of risks. Subscription services (including regular fixed investment plans) are suspended from February 3, 2026, with restoration time to be announced separately. The fund is operating normally, and there is no material information that should have been disclosed but has not been disclosed.

According to ChainCatcher, at the “Crypto 2026: From Cryptocurrency to Intelligent Economy” forum held in Hong Kong, guests including KiteAI Asia-Pacific Director Laughing, Sentient Asia-Pacific Director Anita, Brevis Co-founder and CEO Michael, Mentis Product Director Jerry, and Predict.fun Asia-Pacific BD Director Ah Huang Ricardo participated in a roundtable discussion themed “AI × Crypto — The Foundation of the Next-Generation Digital Economy.” Regarding whether the boom of open-source AI Agents (such as “raising lobsters”) is a short-term sentiment or a long-term trend, Laughing believes this marks a paradigm shift for AI from “conversational” to “executive,” signaling the beginning of a long-term trend. However, as Agents lack real identities, it’s difficult for businesses to trust their transaction behaviors. Issues such as data leakage and uncontrollability remain obstacles that must be overcome for large-scale adoption. Based on his own user experience, Jerry pointed out that current Agents face risks such as memory fuzziness, token-consuming infinite loops, and accidental file deletions. New Agents have shown some improvements but still have shortcomings. Ah Huang Ricardo said that behind the short-term sentiment lies real user demand, but truly profitable and reproducible trading Agents are still very rare, and there is a long way to go. On the core bottlenecks of trust, security, and verifiability, Michael emphasized that the current Agent execution process is a black box—users cannot verify whether decisions are executed as promised. Zero-knowledge proof technology can mathematically prove the origin of an Agent’s output and compliance of its behavior without exposing privacy. Anita added that Sentient is committed to building an open, collaborative AI agent network, promoting in-depth integration of AI and blockchain at the layers of identity, data, and incentives, providing foundational support for the next-generation digital economy. As for the application scenarios most likely to scale first, the guests generally believe that the programming field is already relatively mature. In prediction markets, Agents outperform humans in specific tracks due to their rationality and lack of emotion. They are optimistic in the short term about API Marketplaces and cross-platform consumer shopping Agents; in the medium term about content-creating AI (such as AI influencers); while Agent-to-Agent interaction and Agent/Skill Marketplace are seen as important upcoming trends.