Flash News

TREE NEWS update: China’s State Administration for Market Regulation approved and released five national standardization guiding technical documents covering neuromorphic computing and photonic computing, including a reference architecture for neuromorphic computing and a design guide for optical matrix computing systems. The regulator said the move marks important progress in the country’s standardization layout for future computing.

TREE NEWS update: Changpeng Zhao said on September 15 that it would be “really cool” to see someone create an “immortal fruit fly” on BNB Chain. The former Binance CEO made the remark in a public post, without detailing any specific project, funding or technical plan behind the idea. The comment follows Google’s open-sourcing of a fruit fly brain model that has since been used to play games and trade crypto.

TREE NEWS update: Bill Gates said no government worldwide is prepared for the sweeping social changes AI will bring, warning of risks including job losses, cyberattacks and psychological addiction. In an interview with Reuters, the Microsoft co-founder, who had long been optimistic about AI, said countries including the US and China should work together to manage the technology. He added that if managed well and distributed fairly, AI could help the world’s poorest people.

TREE NEWS update: Bilal Chughtai, a researcher who recently resigned from Google DeepMind, warned that humanity has little time left to prevent widespread harm from AI. In posts on X and LinkedIn on Monday, he said he witnessed AI’s development at Google and is extremely concerned about its current trajectory, adding he genuinely believes AI could kill everyone. He is the latest industry figure to publicly voice such concerns.

TREE NEWS update: ByteDance CEO Liang Rubo said on September 15 at the 2026 Feishu Future Unlimited conference in Beijing that the company has integrated Doubao, Feishu and Volcano Engine to build workplace and productivity products, and will commit more resources to helping AI deliver value inside enterprise organizations. He said agents, models and collaboration environments must be tightly fused once agents become usable.

TREE NEWS reports: China has built more than 70 major computing-power corridors around its national computing hubs as of the end of June, the National Bureau of Statistics said at a State Council Information Office briefing on August’s economic data. Investment in internet and related services tied to the “six networks” rose 42% year on year in the first eight months, while air transport and water transport investment grew 16.7% and 14.7% respectively and power supply investment rose 12.7%.

TREE NEWS update: Hesai Technology announced a new strategic partnership with JD.com to explore physical AI across full scenarios, with Hesai’s 6D full-color lidar and high-precision 3D spatial perception supporting JD’s unmanned logistics vehicles and robots. The two aim to expand physical AI’s real-world applications, Hesai said.

TREE NEWS update: The 2026 Artificial Intelligence Industry Conference will be held in Jinan, Shandong province, from October 14 to 16, under the theme of accelerating AI-enabled new industrialization. The event will showcase AI development results across five directions: independent innovation, application-driven development, ecosystem coordination, open win-win cooperation, and security and controllability.

TREE NEWS reports: Preorders for the Nubia NaviX Ultra, billed as the world’s first AI-agent smartphone and jointly built by ZTE and ByteDance, surpassed 360,000 units as of 9:30 on September 15.com platform data. The device, marketed as the ‘Doubao Phone,’ is the first handset from the two companies’ collaboration.

TREE NEWS reports: The Bank for International Settlements said the AI boom that has driven global equities sharply higher over the past two years is showing growing signs of fragility. In its latest report, the BIS said investors are becoming “increasingly cautious” about the future profitability of AI investment, particularly as leverage at large US technology companies keeps climbing. It also flagged uncertainty from geopolitical tensions and energy price swings, with public finances under pressure.

TREE NEWS update: Cohere CEO Aidan Gomez said AI models are becoming the most powerful cyber weapon ever, citing their ability to discover and exploit vulnerabilities at scale. His warning follows incidents in which AI models breached external systems, including an OpenAI model gaining unauthorized access to Hugging Face. The cybersecurity concerns are feeding a broader AI safety debate, with leaders including Anthropic CEO Dario Amodei calling for measures to slow frontier AI development.

TREE NEWS update: Anthropic’s Claude Opus 5.2 has quietly entered gray-scale testing inside Claude Code, developers found, with request status checks showing the backend model slug now points to Opus 5.2 even though the front-end name is unchanged. Users report that calling Opus 5 in Claude Code performs markedly differently from the web version. A prior internal risk report said Anthropic is deploying an unreleased ‘Model 2’ that writes most of the company’s code and would replace 85% of its research team.

TREE NEWS update: Google has granted all of its engineers access to Claude, the AI model developed by Anthropic PBC. The move makes Anthropic’s model available company-wide to Google’s engineering staff. No further terms, scope or timeline were disclosed.

TREE NEWS reports: Chinese banks have launched their autumn campus recruitment drives with sharply higher demand for AI and large-model specialists. State-owned and joint-stock banks are adding roles in technology elite programs and intelligent R&D, favoring candidates who combine AI skills with financial-business knowledge. The shift comes as banks move large models from pilot testing to scaled deployment, reshaping talent needs and business models.

TREE NEWS update: OpenAI’s top lobbyist, Lehane, met with US lawmakers on Capitol Hill on Monday, and plans further meetings with congressional leadership. The outreach comes as the AI company steps up its engagement with Washington. No specific policy agenda or legislative items were disclosed.

TREE NEWS update: Tempus AI Inc. founder Eric Lefkofsky has backed Anthropic PBC chief executive Amodei’s call to slow the pace of AI model development. Lefkofsky’s support adds a prominent healthcare-AI voice to the debate over whether frontier model releases should be decelerated. No further details or timeline were given.

TREE NEWS update: OpenAI chief Sam Altman warned that humans could lose control of artificial intelligence and called for safeguards during model training. He said the industry should adopt shared standards, arguing developers can act before legislation arrives. Altman did not announce a specific new policy or timeline.

TREE NEWS update: The Trump administration’s AI chief said AI laboratories do not need the government to control the pace of technological development, while the US should have transparency and audit rules for AI labs. The official also said AI labs can develop safety technology without making antitrust concessions.

TREE NEWS update: US President Donald Trump phoned Nvidia CEO Jensen Huang on Monday to discuss artificial intelligence, the White House said. The call marks a direct presidential engagement with the head of the world’s most valuable chipmaker, whose AI processors sit at the center of US export-control and technology policy debates.

TREE NEWS reports: Shield AI is in talks to raise a new funding round at a valuation of at least $20 billion, The Information reported. No round size, investor names or timeline were disclosed. The defense-technology company would rank among the most valuable private AI startups if the deal closes at that level.