Concerns about artificial intelligence are intensifying as researchers warn of potential existential risks. Jacob Coxon, an Anthropic researcher, stated there is over a 10 percent chance AI could kill all humans within the next decade, leading to his resignation. Evan Hubinger from Anthropic agreed with these serious concerns, noting that many builders privately fear the technology could end humanity by 2030.
Professor Krueger reinforced these warnings, citing surveys of over 1,000 experts who believe there is at least a 10 percent chance of an extinction-level event caused by AI. He explained that this scenario involves people losing all authority over the technology, comparing the prospect to science fiction becoming reality.
Despite these risks, the industry continues to advance rapidly. AMD unveiled the Threadripper Halo Station, a personal supercomputer featuring up to two terabytes of memory and 96 cores capable of running AI models with over a trillion parameters. The device is expected to launch early next year with a price likely exceeding 100,000 euros for corporate buyers.
However, the benefits come with significant security costs. A new report from CrowdStrike found that 88 percent of software vulnerabilities were weaponized by attackers within 48 hours in the first half of 2026. IBM research showed that data breaches involving AI now cost an average of $6.04 million, which is about $1 million more than non-AI breaches.
Companies are adapting to these challenges through strategic partnerships. Telefonica and Melia Hotels International signed a new agreement to accelerate digital transformation using AI and cybersecurity, supporting more than 250 hotels across 40 countries. Melia currently handles over 240,000 support requests annually, with more than 70 percent already automated through their existing systems.
Privacy remains a priority as Apple introduces new AI tools like Siri Recap and Live Rewind. These features use always-on microphones but rely on Secure Enclaves to encrypt data and delete raw audio immediately after processing. These features are currently opt-in and will not be available in the European Union due to the Digital Markets Act.
Innovation extends to creative fields as well. ElevenLabs has signed its first deal with Universal Music Group to launch a new AI-powered music platform. This partnership aims to let fans co-create music through remixes while ensuring artists share in the value created, focusing on responsible AI that respects human artistry.
Meanwhile, schools are turning to technology for immediate social needs. More than 200 schools in 19 states are using an AI chatbot named Kiwi to help students with mental health issues due to a shortage of counselors. The chatbot alerts human staff when a student is in crisis, with one principal reporting it helped save a student who had planned to harm themselves.
Finally, the development of secretive advanced AI systems raises further questions. A new system designed to simulate human intelligence is being developed with limited details available to the public. Although the technology shows promise for solving big problems like climate change, experts fear it could surpass human intelligence without proper safeguards or transparency.
Key Takeaways
- Anthropic researcher Jacob Coxon claims there is over a 10 percent chance AI could kill all humans within the next decade.
- Professor Krueger cites surveys of over 1,000 experts who believe there is at least a 10 percent chance of an extinction-level event caused by AI.
- AMD unveiled the Threadripper Halo Station, a personal supercomputer with up to two terabytes of memory and 96 cores.
- Data breaches involving AI now cost an average of $6.04 million, which is about $1 million more than non-AI breaches.
- 88 percent of software vulnerabilities were weaponized by attackers within 48 hours in the first half of 2026.
- Telefonica and Melia Hotels International signed a deal to support more than 250 hotels across 40 countries with AI and cybersecurity.
- Apple introduced Siri Recap and Live Rewind, which use always-on microphones but encrypt data using Secure Enclaves.
- ElevenLabs partnered with Universal Music Group to launch a new AI-powered music platform for fan co-creation.
- More than 200 schools in 19 states use the Kiwi AI chatbot to assist students with mental health issues.
- Experts warn that secretive advanced AI systems developed without transparency pose significant risks to humanity.
Anthropic researcher warns AI could kill humans
An Anthropic researcher named Jacob Coxon claimed there is over a 10 percent chance AI could kill all humans within the next decade. He stated that many AI builders privately fear this outcome because they believe the technology could end humanity by 2030. Another expert, Evan Hubinger from Anthropic, agreed with these serious concerns about losing control of AI systems. These warnings follow a viral post where Coxon said he quit the company due to fears that AI development is too risky.
Professor says humans may lose control of AI
Professor Krueger warned that it is more likely than not that humans will lose complete control over AI technology. He cited surveys of over 1,000 experts who believe there is at least a 10 percent chance of an extinction-level event caused by AI. Krueger explained that this scenario involves people losing all authority over the technology, which he compares to science fiction becoming reality. His comments follow a resignation by Jacob Coxon who feared AI could destroy humanity within a few years.
AMD launches powerful personal supercomputer
AMD unveiled a new device called the Threadripper Halo Station at a tech event in Berlin on Friday. This personal supercomputer features up to two terabytes of memory and 96 cores to run AI models with over a trillion parameters. AMD Senior Vice President Jack Huynh demonstrated its power by creating a 3D world and flight simulator from a single prompt. The device is expected to launch early next year with a price likely exceeding 100,000 euros for corporate buyers.
AI cyberattacks cost millions and spread fast
A new report from CrowdStrike found that 88 percent of software vulnerabilities were weaponized by attackers within 48 hours in the first half of 2026. IBM research showed that data breaches involving AI now cost an average of 6.04 million dollars, which is about 1 million dollars more than non-AI breaches. These findings confirm warnings from over 110 companies that AI makes cyberattacks more widespread and sophisticated. Attackers use AI tools to generate exploit code quickly, shrinking the time defenders have to patch security holes.
Telefonica and Melia expand AI and security deal
Telefonica and Melia Hotels International signed a new agreement to accelerate digital transformation using AI and cybersecurity. This partnership will support more than 250 hotels across 40 countries with cloud services and advanced security measures. The deal builds on a decade of collaboration that already covers 150 hotels in Spain with over 9,000 connected technology devices. Melia currently handles over 240,000 support requests annually, with more than 70 percent already automated through their existing systems.
AI is rapidly changing how mathematicians work
Artificial intelligence is causing a major shift in the field of mathematics, creating discoveries faster than ever before. This progress changes how mathematicians solve problems and train new students, as computers can now handle tasks that once took years. Experts worry that without proper training, future mathematicians might not understand the complex results produced by AI. Additionally, questions arise about who should get credit for discoveries made with the help of these powerful tools.
Apple adds always-on AI features with privacy safeguards
Apple is introducing new AI tools like Siri Recap and Live Rewind that use always-on microphones to process audio on devices. The company uses Secure Enclaves to encrypt data and delete raw audio immediately after processing to protect user privacy. These features are currently opt-in and will not be available in the European Union due to the Digital Markets Act. Security experts suggest users enable two-factor authentication to further protect their iCloud data from potential hackers.
ElevenLabs partners with Universal Music Group for AI music
ElevenLabs has signed its first deal with a major record label, Universal Music Group, to launch a new AI-powered music platform. This partnership aims to let fans co-create music through remixes and personalized vocal experiences while ensuring artists share in the value created. The new platform will be separate from ElevenLabs existing apps and will focus on responsible AI that respects human artistry. Both companies believe this collaboration will unlock new revenue opportunities for the creative community.
Schools use AI chatbots to help with mental health crisis
More than 200 schools in 19 states are using an AI chatbot named Kiwi to help students with mental health issues due to a shortage of counselors. The chatbot, created by the company Alongside, talks to students about their challenges and alerts human staff when a student is in crisis. One principal reported that the system helped save a student who had planned to harm themselves. While some experts worry about the safety of unregulated chatbots, schools view this tool as a necessary aid to expand mental health services.
Experts worry about risks of secretive advanced AI systems
A new AI system designed to simulate human intelligence is being developed in secret with limited details available to the public. Although the technology shows promise for solving big problems like climate change, experts fear it could surpass human intelligence without proper safeguards. The system currently struggles to understand human language nuances, yet its rapid development raises concerns about potential dangers. Critics argue that creating such advanced AI without transparency and safety measures poses a significant risk to humanity.
Sources
- AI researchers 'earnestly believe' it could kill all humans within the next decade, Anthropic researcher says
- Could AI kill humanity? Professor warns we may lose control
- AMD unveils ‘personal super computer’ for new era of computing
- AI Weaponizes Bugs in 48 Hours, Breaches Hit $6.04M
- Telefónica accelerates the evolution of Meliá’s global technology model through AI and cybersecurity
- What's next for mathematics now that AI is upending the field?
- How Apple is trying to normalize always-on AI
- ElevenLabs strikes first deal with a major label to launch AI-powered music platform and audio products
- ‘We actually saved a kid’: Schools recruit AI chatbots as counselor shortage persists
- Behind the Curtain: The AI twilight zone
Comments
Please log in to post a comment.