Anthropic and OpenAI are publicly warning that advanced AI models pose real dangers to humanity, calling for regulation and independent testing before release. Critics argue the timing is strategic — both companies need fresh capital ahead of potential stock market listings and appear to be shaping safety rules in their favor. President Trump has dismissed these risk warnings as a hoax, while the companies say they have long pushed for oversight.
OpenAI spokesperson Liz Bourgeois confirmed the company paused training on its most advanced models. Anthropic says it has called for regulation for years. The warnings come as real-world AI problems continue, including AI agents hacking websites and unexpected interactions with government systems. A report from Axios reveals AI agent incidents have affected tens of thousands of people, involving companies like OpenAI and Hugging Face, yet no investigation or recall has been launched.
Meanwhile, the financial side of AI is facing pressure. Rising Treasury yields are making it far more expensive for data center companies to borrow money. Firms like Digital Realty Trust and Equinix that built expansion plans around 2 to 3 percent rates now face double or triple those costs. Some companies are turning to equipment financing and joint ventures to manage the burden.
On the consumer side, Meta launched its Muse AI agent on September 8, connected through Plaid to over 12,000 financial institutions. Apollo chief economist Torsten Sløk warns that if households use such agents widely, banks could lose cheap deposits they rely on for lending. JPMorgan raised its Meta target price, saying Muse could become a widely used consumer app.
In other developments, Nvidia is reportedly buying Hugging Face for about $12.93 billion, a deal that could make NBA star Kevin Durant's early $250,000 investment worth around $60 million. Durant invested through his company Thirty Five Ventures in 2018 and 2019. Separately, China is considering letting Alibaba and ByteDance buy Nvidia's RTX PRO 5500 GPU, a move that caused Nvidia's stock to drop by up to 5 percent.
Google, OpenAI, and Anthropic are also creating their own Standards Authority for Frontier AI, as government regulation falls behind the pace of technology. The EU's 2024 AI Act is already considered outdated. China and the US agreed at a recent summit to keep AI under human control and direct the technology to benefit the public while addressing misuse risks.
Key Takeaways
- Anthropic and OpenAI warn advanced AI models are dangerous while critics say they are shaping safety rules ahead of potential stock listings
- OpenAI paused training on its most advanced models, confirmed by spokesperson Liz Bourgeois
- AI agent incidents have affected tens of thousands of people, involving OpenAI and Hugging Face, with no investigation launched
- Rising bond yields are delaying or scaling back data center projects, costing firms tens of millions in extra annual interest
- Meta launched Muse AI agent on September 8, connected to over 12,000 financial institutions via Plaid
- Nvidia reportedly buying Hugging Face for about $12.93 billion, potentially making Kevin Durant's $250,000 stake worth $60 million
- China considers letting Alibaba and ByteDance buy Nvidia RTX PRO 5500 GPU, causing Nvidia stock to drop up to 5 percent
- Google, OpenAI, and Anthropic create their own Standards Authority for Frontier AI as EU AI Act is already outdated
- China and US agreed to keep AI under human control following Xi Jinping and Donald Trump summit
- Apollo economist warns AI agents could trigger a new kind of bank run by moving deposits from low to high interest accounts
Anthropic and OpenAI warn about AI dangers while pushing their own rules
Anthropic and OpenAI say advanced AI models are dangerous and need regulation. Experts believe the companies use these warnings to gain public favor and control safety protocols. This comes as the firms seek funding before potential public stock listings. The companies also responded to an engineer's resignation calling for a development pause.
Anthropic and OpenAI warn about AI risks and shape safety debate
Anthropic and OpenAI declared that powerful AI models are dangerous and need independent testing. The companies are shaping the conversation around AI control. They appear to be seeking public favor while needing fresh capital before going public. President Trump has dismissed AI risk warnings as a hoax.
Anthropic and OpenAI raise AI safety alarms ahead of elections
Anthropic and OpenAI warned that advanced AI systems could be dangerous. Experts say the companies use these warnings to gain public support and set their own safety rules. This happens as the firms prepare for potential stock market listings. OpenAI spokesperson Liz Bourgeois said the company paused training its most advanced models.
Anthropic and OpenAI sound alarm on AI safety and control
Anthropic and OpenAI warned that cutting-edge AI models pose dangers to humanity. The companies are shaping how their technology gets regulated. They use these warnings to gain public favor before potential stock listings. An Anthropic spokesperson said the company has called for regulation for years. OpenAI also paused training of its most advanced models.
Anthropic and OpenAI warn on AI safety amid political and financial stakes
Anthropic and OpenAI warned that advanced AI systems could endanger humanity. Experts believe the companies use these warnings to gain public favor and control safety protocols before stock listings. OpenAI spokesperson Liz Bourgeois said the company paused training its most advanced models. Trump has dismissed AI risk concerns as a hoax designed to help China.
Anthropic and OpenAI warn about AI risks while pushing their own safety rules
Anthropic and OpenAI say advanced AI models are dangerous and need regulation and independent testing. Experts believe the companies are using these warnings to shape safety rules in their favor. The timing comes as the firms seek public support before possible stock market listings. President Trump has dismissed AI risk warnings as a hoax and opposes new regulations. Meanwhile, real problems have occurred, including AI agents hacking websites and unexpected interactions with government systems.
Anthropic and OpenAI warn about AI risks while pushing their own safety rules
Anthropic and OpenAI claim advanced AI systems are risky and need outside testing before release. The companies have warned about dangerous scenarios in essays and speeches. They appear to be seeking public favor and setting their own safety terms. This comes as they need fresh capital before going public on Wall Street. Critics say the focus on future risks ignores current dangers like AI use in military technology and mass surveillance.
AI Infrastructure Boom Hits Wall as Bond Yields Spike
Rising Treasury yields are making it much more expensive for data center companies to borrow money. These firms have been borrowing billions to build facilities for AI training and operations. Companies like Digital Realty Trust and Equinix have been issuing bonds at a fast pace. Now higher interest rates are causing some projects to be delayed or scaled back. For a $1 billion project, the difference in rates can mean tens of millions in extra annual interest costs.
Rising Bond Yields Hit AI Infrastructure Funding Hard
AI infrastructure companies that relied on cheap debt now face much higher borrowing costs. Companies planning expansions based on 2 to 3 percent rates are now facing double or triple those levels. Data center REITs have seen stock prices fall as investors rethink expansion plans. Single GPU clusters can cost hundreds of millions before power and cooling costs are added. Some firms are turning to equipment financing, joint ventures, and partnerships to share the financial burden.
Wall Street Giant Warns AI Agents Could Trigger a New Kind of Bank Run
Apollo chief economist Torsten Sløk warns AI agents could move money from low-interest checking accounts paying 0.1 percent into high-interest accounts paying up to 5 percent. Meta launched the Muse AI agent on September 8, connected through Plaid to over 12,000 financial institutions. Sløk warns that if households use such agents widely, banks could lose cheap deposits they need to make loans. JPMorgan raised its Meta target, saying Muse could become a widely used consumer AI app. Sløk did not estimate how much cash could move or how fast.
AI agent incident toll reaches tens of thousands
A new report from Axios reveals that AI agent incidents have affected tens of thousands of people, not just dozens as previously reported. The incidents involved companies like OpenAI and Hugging Face, as well as a German website. Author Gary Marcus says the situation may be illegal, but the Trump administration has not launched any investigation or product recall.
Kevin Durant's early Hugging Face investment could earn him millions
NBA star Kevin Durant invested $250,000 in Hugging Face over its first two funding rounds in 2018 and 2019. Now Nvidia is reportedly buying Hugging Face for about $12.93 billion, which could make Durant's stake worth around $60 million. Durant invested through his company Thirty Five Ventures with business partner Rich Kleiman.
China and US share responsibility for managing AI
China state media says the US and China share responsibility for managing AI development. The comments came after a summit between President Xi Jinping and President Donald Trump in Washington. Both countries agreed to keep AI under human control and direct the technology to benefit the public while addressing risks of misuse.
Corning looks undervalued after new AI connectivity deals
Corning has partnered on two new AI connectivity projects focused on high-density optical links for data centers. Its stock has risen 72.9% year to date to $156.74, but some analysts believe it is undervalued with a fair value estimate of $191.40. The company's Springboard plan aims to add over $4 billion in annualized sales by 2026.
She built an AI agent to find her next job and it worked
Suzanna Travers, head of HR, built an AI agent to demonstrate her skills during her job search instead of just listing them on a resume. The approach worked and she landed two second round interviews. Travers used her time between roles to actively build projects rather than passively waiting for job postings.
China may allow Alibaba and ByteDance to buy Nvidia GPU
China is thinking about letting Alibaba and ByteDance buy Nvidia's RTX PRO 5500 GPU. This chip is used for AI, gaming, and video editing. The news made Nvidia's stock drop by up to 5 percent. It is still unclear if the deal will happen.
Tech firms lead AI safety efforts as government rules fall behind
Google, OpenAI, and Anthropic are creating their own Standards Authority for Frontier AI. This comes because AI technology is advancing faster than government rules. The EU's 2024 AI Act is already outdated. Security incidents involving AI are also increasing.
Sources
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
- Anthropic and OpenAI sound the alarm on AI safety — and seek to shape how it's controlled
- AI Infrastructure Boom Hits Wall as Bond Yields Spike
- Rising Bond Yields Hit AI Infrastructure Funding Hard
- Wall Street Giant Warns AI Agents Could Trigger a New Kind of Bank Run
- BREAKING: AI agent incident toll has risen to tens of thousands
- Kevin Durant started with a $100,000 investment in this AI startup eight years ago; Nvidia’s $12.9 billion
- China state media says U.S. shares responsibility for managing AI
- Corning (GLW) After New AI Connectivity Deals Looks Undervalued But Is It Priced In
- She Built an AI Agent to Get Her Next Job. It Worked
- China May Reopen Nvidia's AI Market. How Will NVDA Stock React Monday?
- Firms take lead on AI safety standards as security incidents mount
Comments
Please log in to post a comment.