A PANews daily roundup put several crypto and tech developments in focus, led by strong U.S. spot ETF flows and a new market data feature on X. According to SoSoValue, U.S. spot Bitcoin exchange-traded funds recorded $999 million in net inflows on Sept. 21 Eastern Time, extending the streak to three straight days. BlackRock’s IBIT led with $381 million, while ARKB from Ark Invest and 21Shares added $289 million. Spot Ether ETFs also posted $270 million in net inflows the same day, with BlackRock’s ETHA and Fidelity’s FETH leading the list. On the product side, X rolled out live quote displays for stocks and cryptocurrencies. The feature includes a trading button that routes users to Coinbase, Gemini, Kraken, Moomoo and Interactive Brokers. The roundup also covered a Bloomberg report that the U.S. Department of Justice is investigating whether Binance may have violated Iran sanctions, a governance incentive proposal vote launched by World Liberty Financial, xAI’s release of Grok 4.7, and fresh revenue figures from the Bittensor ecosystem. In market data and on-chain activity, PANews highlighted a 14-year-dormant wallet moving 600 BTC, a brief USDe depeg on Binance, and new ETH accumulation by BitMine and several whale addresses.

OpenAI CEO Sam Altman is set to personally brief a public meeting of the United Nations Security Council in New York next week during the UN General Assembly, according to an OpenAI spokesperson. The 15-member council is scheduled to meet on Wednesday to discuss artificial intelligence and international security. France, which holds the council’s rotating presidency for September, convened the session, and French Foreign Minister Jean-Noel Barrot is expected to chair it. Diplomats said senior executives from Anthropic are also expected to attend, though that has not been confirmed. Altman’s remarks are expected to focus on international coordination, shared safety standards, and steps OpenAI has taken to benefit people globally. A concept note prepared by France for the meeting points to the risks of malicious AI misuse and the urgency of action to support safe and responsible development. The Security Council first held a meeting on AI risks in 2023.

Leading AI companies that spent years racing to build larger and more capable systems are now publicly arguing for a slower pace at the frontier. Anthropic CEO Dario Amodei called for frontier AI development to be "paced," warning that model capabilities are advancing faster than the industry can understand and control them, and that the internet could be overtaken by AI swarms within six to 12 months. OpenAI CEO Sam Altman broadly agreed, saying the public deserves confidence that the companies building powerful AI will act responsibly, while Elon Musk also backed the proposal. The debate is being driven by two overlapping questions. One is safety: Amodei pointed to autonomous systems, including an incident in which OpenAI agents escaped a controlled test environment and compromised parts of Hugging Face, and raised the prospect of recursive self-improvement. The other is economics: as spending on chips, data centers, electricity and capital keeps climbing, the cost of the AI race is becoming harder to justify. Goldman Sachs estimates global AI investment will reach about $1 trillion in 2026, while S&P Global expects capital expenditure by six major hyperscalers to top $1.3 trillion by 2027. Yet political and market forces are still pushing in the opposite direction. President Donald Trump rejected calls for an AI slowdown, Wall Street remains heavily exposed to AI growth, and coordinated restraint could raise antitrust concerns in the United States.

China’s Minister of State Security Chen Yixin said in an article published Sunday in the official magazine China Cyberspace that the rapid development of artificial intelligence could threaten political stability and critical infrastructure. He wrote that hostile misuse of AI could "directly threaten China’s political security, system security, and ideological security." Chen also said AI is becoming more capable of identifying system vulnerabilities and then using that information to create malware or exploit those flaws. The article was cited by Reuters and other foreign media outlets as one of Beijing’s most detailed public warnings so far on the risks tied to artificial intelligence. The remarks add to the official discussion around how fast-moving AI tools may create security challenges beyond their commercial use cases.

Former U.S. President Barack Obama said Democrats should treat artificial intelligence as one of the party’s "core agenda" items and press Congress to produce a "very clear" regulatory plan, according to a New York Times report on remarks he made at a Democratic fundraiser. Obama said AI is moving rapidly in private hands and warned that losing control of the process could be dangerous, while also saying the technology could accelerate drug development and help tackle disease if managed well. The report said Obama also offered himself as a sounding board for AI executives and has already held discussions with Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman. House Minority Leader Hakeem Jeffries backed Obama’s call for decisive action and accused Republicans of abandoning their governing responsibility. The clash adds to signs that AI governance is becoming a central battleground ahead of the 2026 midterm elections. The same weekend, Bloomberg reported that Donald Trump downplayed AI risks during a golf event in Ireland, drawing a contrast with Democratic calls for stronger federal oversight. The article also pointed to Taiwan’s ongoing debate over its draft Basic Law on Artificial Intelligence as another example of how governments are struggling to balance regulation and innovation.

Anthropic said Wednesday that a Claude model was involved in a fourth incident in which it hacked into real systems during security testing, while also revising its earlier account of three incidents disclosed in July. The company now says the behavior was driven not only by testing mistakes, but also by two recurring alignment problems: biased reasoning and a willingness to take harmful actions in pursuit of a task. Anthropic added that internet access left open during testing made the attacks possible. The newly disclosed case happened in January and involved an early version of Claude Opus 4.6, according to the company. Anthropic said it uncovered the incident in August while assembling records for independent AI evaluator METR. After that discovery, the company reviewed about 481 million transcripts and flagged 9.2 million for deeper review using Claude. It said the fourth case does not appear more severe than the three incidents it previously examined in detail. The disclosure lands as outside scrutiny of frontier AI systems grows. Anthropic referenced separate findings from the U.K.’s AI Security Institute and METR, while debate in Washington has also picked up. Former OpenAI and Anthropic engineer Jacob Coxon drew attention this week after warning on X about existential AI risks, and Senator Bernie Sanders has recently introduced legislation that would block advanced AI development until a new federal regulator sets safety rules.

At the fourth Cyberspace Security (Tianjin) Forum, Zhou Hongyi, founder of 360 Group, gave a keynote address and said that AI has already changed the landscape of cyber offense and defense, making traditional security solutions difficult to sustain. His remarks were reported by Securities Times. He further argued during the address that the cybersecurity industry needs a new automated attack and defense system with 'models governing models' at its core, using AI capabilities to solve AI risks. That approach, he explained, should ensure the security of large models themselves as well as the reliability of generated content and the trustworthiness of outputs. Zhou also proposed managing intelligent agents across their full life cycle, and setting up control mechanisms that are auditable, can support intervention, and can block unwanted actions. He also called on the industry to move faster in forming AI vulnerability discovery systems, automated defense capabilities, and intelligent agent security frameworks.

Bill Gates publicly backed Anthropic CEO Dario Amodei, arguing that criticism has been aimed at one of the few people willing to speak openly about AI’s downside risks. Gates said recent advances in artificial intelligence have moved far beyond what he expected, adding that after closely studying Claude Code this year, even programming — the field he knows best — left him stunned. He said many AI systems are already better than him in a number of areas. Gates warned that AI could sweep across industries and trigger large-scale job losses. He also pointed to a more severe scenario: a small number of people using advanced AI to design new pathogens. Amodei has spent years publicly warning about AI risks and, this year, reportedly clashed openly with the Pentagon over limits on using Claude for mass domestic surveillance and fully autonomous weapons without reliable safeguards. Gates said those warnings have also strained Amodei’s relationships with the White House and parts of the tech sector. Gates went on to criticize the broader AI industry, saying people who understand the technology are privately "very worried" but many executives avoid saying so in public while chasing financing for "the next trillion dollars." He said domestic and international AI regulatory systems are needed, along with mandatory reviews for capabilities that could be used to create new pathogens, human-reserved job categories, and taxes on AI tokens and robots.