Tech leaders are divided on how fast AI development should move. Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman favor coordinated slowdowns with government rules. Meta CEO Mark Zuckerberg thinks labs should set their own pace while moving safely. NVIDIA CEO Jensen Huang pushes back on new regulations, arguing the market already handles safety.
Al Gore says the public backlash against AI data centers is driven more by job loss fears than by carbon emissions. He views warnings from AI leaders at OpenAI and Anthropic as sincere, though he is concerned some companies are using methane gas turbines and favors renewables and batteries.
On the military side, critics told Congress that AI-driven targeting can outpace human approval. Anna Mysyshyn warned that a person managing multiple AI systems may miss errors, while Amanda Klasing said AI shortens the kill chain and makes accountability harder. Joseph Chapa called for more AI training and better auditing after harm occurs.
A separate position paper argues language models should not guide military planning or crisis response without strong safety checks. It identifies five failure modes, including decision laundering and escalation through AI adjudication, and says wargames should test weaknesses rather than prove systems are safe.
Elsewhere, former US envoy for AI diplomacy doubts safety talks with China will succeed, expecting Beijing to downplay safety concerns and use cooperation as a bargaining tool. In South Carolina, a Senate committee held its first AI hearing on September 16, with nine lawmakers studying AI risks and aiming to release a report by September 2028.
Lectra launched Apogy, a cloud-based agentic AI tool for fashion development that helps teams move from idea to prototype using 3D simulations and realistic renderings, with early customers including O'Neills and Oniverse. ASUS released a report urging companies to move beyond AI pilots to full AI factories, stressing that success requires data, networking, storage, power, cooling, and operations, not just GPUs.
Jeff Reed, a former software engineer and linguist, is using AI to decode wolf sounds in Yellowstone. His roughly 50 recording devices collect 500,000 hours of sound per year, and his AI tools pull out over 8,000 hours of wolf calls, creating the world's largest wolf sound database.
Key Takeaways
- Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman support coordinated AI slowdowns with government rules.
- Meta CEO Mark Zuckerberg says labs should move safely at their own pace.
- NVIDIA CEO Jensen Huang opposes new AI regulations, saying the market handles safety.
- Al Gore says AI job fears drive data center backlash more than carbon emissions.
- Critics warn AI military targeting may outpace human approval and accountability.
- A position paper says language models should not guide military wargaming without strong safety checks.
- Former US envoy doubts AI safety talks with China will focus on real risks.
- South Carolina Senate committee begins AI hearings, aims to report by September 2028.
- Lectra launches Apogy, an AI tool for fashion development with 3D simulations.
- ASUS urges companies to build full AI factories beyond GPU pilots.
Tech leaders split on slowing AI development
Major tech leaders disagree on whether AI development should slow down. Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman support coordinated slowdowns with government rules. Meta CEO Mark Zuckerberg says labs should move safely at their own pace. NVIDIA CEO Jensen Huang opposes new regulations, saying the market handles safety on its own.
Al Gore says AI job fears matter more than data center emissions
Al Gore says the public backlash against AI data centers comes more from fears of job loss than from carbon emissions. He believes warnings from AI leaders at companies like OpenAI and Anthropic are sincere. Gore does worry that some companies are using methane gas turbines, but he prefers renewable energy and batteries.
AI helps researcher decode Yellowstone wolf sounds
Jeff Reed, a former software engineer and linguist, is using AI to study wolf vocalizations in Yellowstone. He built about 50 recording devices that collect 500,000 hours of sound each year. His AI tools pull out over 8,000 hours of wolf calls, creating the world's largest wolf sound database to help understand wolf communication.
Lectra launches AI tool for fashion development
Lectra introduced Apogy, a cloud-based agentic AI tool for fashion product development. It helps teams work together from idea to prototype with 3D simulations and realistic renderings. Early customers include O'Neills and Oniverse, which say the tool improves teamwork and helps attract younger talent.
Former US envoy doubts AI safety talks with China will work
The former U.S. State Department envoy for AI diplomacy says past talks with China failed to focus on AI safety risks. He expects China to downplay safety concerns in future meetings and use cooperation as a bargaining tool. He believes China will continue trying to gain an advantage while portraying itself as a helper for developing nations.
Critics warn AI military targeting may outpace human approval
Critics told Congress that AI-driven military targeting can move faster than humans can verify or stop. Anna Mysyshyn said a person overseeing multiple AI systems may not notice errors in time. Amanda Klasing warned that AI can shorten the kill chain and make accountability harder. Joseph Chapa called for more AI training in the military and better ways to audit AI decisions after harm occurs.
AI not ready for strategic wargaming, position paper argues
A new position paper says language models should not guide military planning or crisis response without strong safety checks. It warns that open-ended AI wargames can create false decisions and hide mistakes. The authors identify five failure modes, including decision laundering and escalation through AI adjudication. They say wargames should be used only to test weaknesses, not as proof that AI systems are safe.
Senate committee begins AI hearings in South Carolina
A South Carolina Senate committee held its first meeting on artificial intelligence on September 16. Nine lawmakers will study AI uses and risks and aim to create state policies and rules. Experts discussed how to define AI, machine learning, and generative tools. The committee plans to keep meeting and hopes to release a full report by September 2028.
ASUS report urges AI factories for enterprise scale
ASUS says companies must move beyond AI pilots to full AI factories for real-world use. The report says success depends on more than GPUs and requires good data, networking, storage, power, cooling, and operations. Priorities differ by region, with some markets focusing on reliability and others on latency or security. ASUS works with partners to help businesses build production-ready AI infrastructure.
Sources
- Divisions emerge in the tech industry over calls for a coordinated AI slowdown
- Al Gore has a surprisingly calm take on the AI data center backlash
- ‘Don’t shoot. Listen’: The AI-aided quest to understand Yellowstone’s wolves
- Lectra Launches Agentic AI for Fashion Product Development
- Opinion | I Led A.I. Diplomacy for the U.S. The Coming Safety Talks Will Not Save Us.
- AI military targeting may move faster than humans can authenticate, critics warn
- Position: AI Is Not Ready for Strategic Conflicts
- Senate committee holds 1st of many meetings on artificial intelligence
- From AI Pilots to AI Factories: Building the Infrastructure for Enterprise Impact
Comments
Please log in to post a comment.