Anthropic researchers, including former employee Jacob Coxon and Alignment Science Lead Evan Hubinger, have issued stark warnings that AI technology could lead to human extinction by 2030. Hubinger specifically stated he believes there is more than a ten percent chance of this outcome within the next decade, a view that has finally drawn attention from US Congress after years of legislative inaction.
These safety concerns extend beyond theoretical risks to immediate threats. Anthropic released a report linking Russian hackers to the use of its Claude AI chatbot to create software for autonomous kamikaze drones targeting Ukraine. The report details how bad actors utilized the tool to build software allowing drones to select targets and coordinate swarms without human control, alongside cyber-attacks on government sectors.
In response to these dangers, Dario Amodei, a co-founder of Anthropic, proposed a three-step plan to slow down the development of top AI models. His strategy involves outside monitors, agreed-upon safety standards among firms, and global coordination. While OpenAI CEO Sam Altman agrees on the need to pace the frontier, the Trump administration currently favors rapid growth, creating a tension between safety and speed.
Despite these high-stakes debates, the industry continues to expand globally. Over 120 startups showcased AI innovations at the 2026 Inclusion Conference in Shanghai, attracting over 500 investors who are increasingly directing funds toward AI projects. Meanwhile, in the US, a free AI roadshow arrived in Indiana to help local employers boost productivity, noting that more than 17% of businesses there already use the technology.
Criticism also arises from the creative sector and privacy advocates. Director Werner Herzog condemned the 2025 film "About a Hero" for using AI to imitate his work and deepfake his narration. Simultaneously, experts warn users to be cautious with ChatGPT, as privacy policies from major developers like OpenAI, Google, and Microsoft reveal they use customer conversations to train models indefinitely.
Key Takeaways
- Anthropic researchers Evan Hubinger and Jacob Coxon warn of a greater than ten percent chance of human extinction by 2030 due to AI risks.
- Anthropic released a report confirming Russian hackers used Claude AI to build autonomous kamikaze drones targeting Ukraine.
- US Congress is finally acting on AI safety legislation following warnings from Anthropic and previous figures like Geoffrey Hinton.
- Dario Amodei proposed a plan to slow AI development, including outside monitors and global coordination, which OpenAI CEO Sam Altman supports.
- Over 120 startups presented AI products at the 2026 Inclusion Conference in Shanghai, attracting over 500 international investors.
- More than 17% of Indiana businesses currently use AI to advance their operations, according to a local roadshow report.
- Director Werner Herzog criticized the 2025 film "About a Hero" for using AI to imitate his style and deepfake his narration.
- Privacy experts advise caution with ChatGPT after finding that OpenAI, Google, and Microsoft retain user data indefinitely for training.
- Director of National Intelligence Jay Clayton highlighted AI as a critical national security risk in the 2026 Annual Threat Assessment.
- Some investors plan to increase their AI investment to over ninety percent of their funds following the Shanghai conference.
Anthropic Researchers Warn AI Could End Humanity by 2030
Jacob Coxon, a former researcher at Anthropic and OpenAI, resigned on September 9, 2026, to warn that AI developers believe their technology might kill everyone by the end of the decade. Evan Hubinger, Anthropic's Alignment Science Lead, agreed with this concern and stated he believes there is more than a ten percent chance of human extinction within the next ten years. These warnings highlight a strange contradiction where companies continue to invest heavily in AI despite fears that it could destroy civilization. The article explains that these experts are discussing a possibility, not a guaranteed event, and notes that the technology is still developing.
Anthropic Report Links Russian Hackers to AI-Powered Kamikaze Drones
Anthropic released a report stating that a small team of Russian developers used its Claude AI chatbot to create software for autonomous kamikaze attack drones. This software allowed the drones to select their own targets and coordinate swarms without human control, specifically targeting the Donetsk region in Ukraine. The report also found that hackers used AI to launch cyber-attacks against Ukrainian government and military sectors, including phishing and wifi hijacking operations. These findings suggest that bad actors are increasingly using AI tools to build dangerous weapons and execute complex cyber-attacks.
Over 120 Startups Showcase AI Innovations at Shanghai Conference
More than 120 startups presented advanced AI products at the 2026 Inclusion Conference held on the Bund in Shanghai from Wednesday to Saturday. The event, themed 'Building the AI Economy Together,' attracted over 500 investors from around the world to view innovations in healthcare, energy, and new materials. Company founders like Wang Yi of Agiphant Tech discussed plans to sell products like the Omniband in North America and Europe. Investors expressed strong interest in projects with clear commercial paths, with some planning to increase their AI investment to over ninety percent of their funds.
US Congress Finally Acts on AI Safety After Years of Inaction
After a decade of failed bills and several high-profile resignations, Washington is finally responding to warnings about artificial intelligence safety. This week, Evan Hubinger from Anthropic publicly stated he believes there is a greater than ten percent chance of human extinction within a decade, which prompted a political reaction. Previous warnings from figures like Geoffrey Hinton and Jan Leike did not result in new legislation, but the current situation has accelerated congressional attention. While federal laws are still being discussed, some states like New York and Colorado have already passed their own AI regulations.
Dario Amodei Proposes Slowing AI Development to Ensure Safety
Dario Amodei, a co-founder of Anthropic, proposed a three-step plan to slow down the development of top AI models to prioritize safety precautions. His plan includes bringing in outside monitors to check for hazards, agreeing on safety standards among AI firms in democratic countries, and achieving global coordination with nations like China. Amodei argues that racing to build more powerful models increases the risk of things going awry, such as when an AI model causes a company to shut down its operations. OpenAI CEO Sam Altman agreed with the need to pace the frontier, though the Trump administration currently favors rapid AI growth.
DNI warns AI poses national security risks
Director of National Intelligence Jay Clayton discussed national security risks on Fox News. He warned that leadership in emerging technologies defines global power. Clayton highlighted the dangers artificial intelligence presents to the United States. This discussion was part of the 2026 Annual Threat Assessment.
Free AI roadshow helps Indiana employers
A free AI roadshow arrives in the South Bend-Elkhart region on September 18, 2026. The event is a collaboration between the Indiana Chamber of Commerce and Notre Dame's iNDustry Labs. Local employers can learn how artificial intelligence boosts productivity and competitiveness. The program includes breakfast, networking, and hands-on sessions starting at 8:30 a.m. More than 17% of Indiana businesses already use AI to advance their operations.
Werner Herzog criticizes AI film imitation
Director Werner Herzog criticized filmmakers who use AI to imitate his work. He spoke out about the 2025 movie "About a Hero," which used an AI-generated script trained on his films. The movie also featured an AI-deepfaked version of Herzog as the narrator. Herzog called the project phenomenally stupid and stated he would not take part in it. He believes AI should replace talentless hacks who only repeat existing styles.
Experts warn against sharing data with ChatGPT
Experts advise users to be careful about what they share with AI chatbots like ChatGPT. A survey from Elon University found that large language model adoption is astounding among the public. Researchers analyzed privacy policies from six major developers including OpenAI, Google, and Microsoft. They found that all these companies use customer conversations to train their models and some keep the data indefinitely. Cybersecurity expert George Kamide warned that personal identifiers and sensitive data could become part of a dataset beyond the user's control.
Sources
- Will Artificial Intelligence Kill Us All by 2030?
- Ukraine war briefing: Russian developers used AI to build ‘kamikaze’ attack drone software, Anthropic says
- Cutting-edge AI products shine at 2026 Inclusion Conference in Shanghai
- After a decade of failed bills and three years of resignations, Washington finally discovers it cares about AI safety
- Dario Amodei: ‘We Owe It to Humanity’ to Slow Down AI
- DNI warns of artificial intelligence national security risks
- Local employers to explore practical AI applications at Notre Dame
- Werner Herzog Demolishes Talentless Hacks Who Use AI to Imitate Other Filmmakers
- Experts Warn You Should Never Share These 5 Things With ChatGPT — And What You Can Do If You Already Did
Comments
Please log in to post a comment.