Zhipu ZCode is officially open-sourced, with its official stating that it has completed security rectification and passed audits conducted by two institutions.

From Beating AI Express, an engineer who recently joined a large company posted on X that his team now relies on Claude Code for nearly all engineering tasks, including specification documents, code, testing, PRDs, work orders, and issue resolutions. He said management views "pushing code as not the bottleneck", so it has repeatedly demanded the team speed up deliveries. Employees work 12 to 13 hours daily, only repeatedly "hitting enter". The engineer added that almost no team members have time to review the generated code or actually fix bugs; from junior to senior engineers, everyone is stuck in a cycle of interacting with AI and submitting code, "no one is thinking anymore". He clarified he does not oppose using AI for programming, but at minimum, time should be allocated to reviewing code and understanding the system. Currently, the team’s sole goal is to launch as quickly as possible, "no matter what". The post was widely shared and sparked heated debate, garnering over 2.42 million views and 22,000 likes on X.

According to Xinhua News Agency, local time on September 20, He Lifeng, China’s lead negotiator for China-US economic and trade talks and Vice Premier of the State Council, held economic and trade consultations in New York with US counterparts: US Treasury Secretary Bessent and Trade Representative Greer. Guided by the important consensus of the two heads of state, and adhering to the principles of mutual respect, peaceful coexistence and win-win cooperation, the two sides held candid, in-depth and constructive exchanges on major economic and trade issues of common concern, including implementing the consensus from previous rounds of economic and trade consultations and promoting bilateral trade and investment, and also held a dialogue on artificial intelligence-related issues.

The cryptocurrency market continues its recovery, with Strategy — the world’s largest corporate Bitcoin treasury firm — now seeing its unrealized profit on Bitcoin holdings expand to $5.02 billion, after previously posting an unrealized loss of $10 billion. Strategy currently holds 845,000 BTC and $6.4 billion in reserve assets.

According to HTX market data, Bitcoin briefly dipped below $81,000, with a 0.09% decline over the past 24 hours.

Beating AI Express (from Dongcha) reports that the AI wave is disrupting the traditional asset diversification logic of pension funds and sovereign wealth funds, with risks spreading from tech stocks to private equity, corporate bonds, infrastructure and other sectors, prompting institutional investors to reassess their overall portfolio AI exposure. Goldman Sachs estimates that AI infrastructure-related companies account for around 40% of the total market capitalization of the S&P 500. Apollo data shows that this year, AI-related issuances make up nearly half of investment-grade bond issuance and 87% of venture capital funding. Monte Tarbox, chief investment officer of the New York City Retirement Systems, recently even rejected a private equity fund’s capital-raising request due to its overexposure to AI holdings. A key challenge facing institutions now is the lack of a unified standard for measuring AI exposure. The Los Angeles County Employees Retirement Association (LACERA) estimates that 8% to 19% of its holdings are AI-related. Invesco’s survey of 90 sovereign wealth funds found that more than half list market concentration as the top risk of AI investments. To address AI risks, some large institutions are adopting the "Total Portfolio Approach (TPA)", breaking down barriers between asset classes such as stocks, bonds, private equity and infrastructure to track overall AI-related exposure and inter-asset correlations. At the same time, some institutions are using AI tools to monitor their own portfolios, avoiding excessive concentration risks while chasing AI gains.

Swing trader 0x4cee, who sold 10,500 $ETH ($23.64M) at an average price of $2,252 a month ago for a $3.66M profit, just bought back 7,567 $ETH ($19.93M) at a higher price 2 hours ago!