AI security startup Alice secures $140 million in funding, with its annual recurring revenue (ARR) approaching $100 million.

Insight Beating AI Flash News: Google Cloud has simultaneously launched Gemini Enterprise for Financial Services and Gemini Enterprise for Legal. These are industry-specific agent platforms built on Gemini Enterprise, marking the first two such versions, with healthcare, life sciences, and other professional services versions also in development. The financial services version targets capital markets and corporate banking. It comes with a built-in Financial Research Agent, over 50 financial capabilities, and 13 data connectors that directly integrate data from FactSet, Moody’s, MSCI, S&P Global, SEC EDGAR, etc., for use in KYC, credit research, portfolio risk analysis, and bond issuance documentation. Deutsche Bank participated in designing the Financial Research Agent and plans to deploy it in its corporate banking division; CME Group has already started using this financial version of Gemini Enterprise. The legal version connects to legal tools and databases including Harvey, Legora, Thomson Reuters, iManage, RelativityOne, supporting contract review, legal research, citation verification, and regulatory tracking. Original system file permissions and case isolation rules will remain in effect, and customer data will not be used to train Google’s base models. This launch is essentially adding "industry suites" to Gemini Enterprise: the model is just the foundation, with core offerings being industry data, permissions, workflows, and specialized agents. Google aims to capture the central entry point for enterprises to access various AI tools.

The U.S. Department of Justice announced that a San Francisco federal jury has found Japheth Dillman, founder of cryptocurrency fund Block Bits Capital, guilty of wire fraud and conspiracy. Prosecutors alleged that between June 2017 and August 2018, Dillman raised nearly $1 million from over 20 investors, claiming the fund used its proprietary trading software "Autotrader" to trade cryptocurrencies and generate returns—though he knew the software was incomplete and non-operational. Dillman and an accomplice subsequently diverted the investment funds to personal expenses and high-risk crypto projects, and after those investments incurred substantial losses, he continued to falsely inform investors the fund had earned significant profits. Dillman is currently released on bail, with sentencing scheduled for December 8; he faces up to 20 years in prison and a $250,000 fine on each count.

According to market data from BIT (Bit.com), U.S. crypto-themed stocks have extended their gains. The performance breakdown is as follows: MSTR rose 2.98%; COIN gained 3.69%, CRCL climbed 3.72%; SBET increased 2.36%; BMNR jumped 4.87%; HOOD advanced 6.20%; and HYPE's crypto treasury firm PURR rose 8.79%.

While Nvidia’s upcoming earnings report and remarks from Federal Reserve Chair Walsh at the Jackson Hole Symposium are the main events capturing investor focus this week, stock derivatives market traders have already begun preparing for potential volatility spikes around the November U.S. midterm elections. Volatility traders closely monitoring futures markets linked to the CBOE Volatility Index (VIX) point to signs that demand for hedging S&P 500 index fluctuations is rising ahead of and around the election. The September-expiring VIX futures are currently trading at around 17.4, while October contracts have climbed to 19, and November contracts have risen further to 19.7. Matthew Thompson, co-portfolio manager at Little Harbor Advisors, said: "The U.S. general election is approaching, and we have entered a time window where elections will impact the VIX. You can already see this 'steepening' in the VIX futures term structure." A study by analysts at CBOE Global Markets shows that since 1945, 80% of midterm election years have recorded higher realized volatility than the prior year, with an average increase of 3.5 volatility points. In years when the same party controls both the White House and Congress, realized volatility rises by an average of 6 volatility points.

News Brief from Dong察 Beating AI: OpenAI has released the latest test results of its self-developed AI inference chip Jalape?o. The chip outperformed NVIDIA’s GB200 and GB300 superchips in the InferenceX benchmark test. On models including GPT-OSS 120B, DeepSeek R1, and Kimi K2.5 1T, Jalape?o delivers 1.5 to 1.9 times the AI workload per watt of competing products, with end-to-end latency reduced by 1.7 to 3.6 times. OpenAI noted that Jalape?o uses an ASIC architecture developed in partnership with Broadcom, purpose-built for AI inference to boost throughput and lower latency. The company plans to roll out small-scale deployments of the chip by the end of this year, expand production in 2027, and continue developing its second and third-generation chips. However, OpenAI stressed that it will not fully replace chips from partners such as NVIDIA.

Beating AI Express: OpenCode co-founder Dax Raad shared a screenshot of Stripe’s backend, revealing that over the past few days, Stripe has blocked roughly $323 million in fraudulent payment attempts on behalf of OpenCode. Dax noted this amount is far higher than OpenCode’s own revenue, adding that if these transactions had been processed, they could have directly driven the company to collapse. OpenCode has been waging a battle against this group of fraudsters: some register thousands of Go accounts at once, use rotating proxies, then resell the account balances via tokens. On August 11, Dax announced he had just purged 7,013 fraudulent accounts, a move expected to save the firm $400,000 monthly. The cancellation of the first-month $5 discount is now viewed as part of the same anti-abuse offensive and defensive efforts.