OpenAI's cybersecurity team has demonstrated how AI agents can identify software vulnerabilities at an unprecedented pace. In a single 30-hour run against the latest Linux kernel, the system discovered 948 local privilege escalation exploits and reproduced 1,724 total vulnerabilities. The operation cost $1.21 million in API credits, roughly $1,279 per exploit, and even uncovered 10 new exploitation techniques never seen before. Human reviewers helped patch 61 of the most serious flaws within 10 days of the mid-October 2026 test.
The findings highlight a growing concern around exploit development timelines. What once took weeks now happens in hours, with the time from vulnerability disclosure to working attack dropping to as little as one day in 2024. Open-weight models can exploit bugs for about $1,000 each, while median fix times for critical vulnerabilities still sit around 30 to 43 days. Patching simply has not kept up with the speed AI enables.
Beyond security research, AI is making its way into consumer products and workplace tools. AT&T's Connected Life home security system became the first to integrate with Google Home Premium through Gemini for Home. The feature adds camera descriptions, natural-language video search, and daily activity summaries, though it requires AT&T's professional monitoring plan at $29.99 per month. AT&T launched Connected Life in December 2025 alongside Google and Abode.
Workplace productivity numbers tell a more nuanced story. AI assistance raised customer-support output by 15% overall and about 30% for less experienced workers. A two-month employee with AI help matched the performance of a six-month veteran, effectively closing roughly three-quarters of the education-based productivity gap. Researchers caution, however, that these gains may flow mostly to capital owners, potentially widening inequality between labor and wealth.
On the education and hiring front, experts say job seekers should prepare for AI-related questions during interviews. Lisa Dubler of Georgetown University advises candidates to discuss how they currently use AI and how they plan to in future roles. ZipRecruiter economist Nicole Bachaud notes that only about 10% of workers highlight AI skills on their resumes. Meanwhile, the Global University Presidents and Leaders Summit 2026 in Hong Kong ended with a joint declaration from over 50 universities across more than 20 countries, stressing the importance of liberal arts education alongside AI training.
SAP research adds context: 69% of organizations say AI expands employee duties, but upskilling has not kept pace. Entry-level workers now need judgment, trustworthiness, and verification skills more than ever. Little Friends, a human services provider, echoes this balance, noting that while 49% of U.S. adults use chatbots, AI cannot replace the human touch required in care work. The central question remains whether AI genuinely improves outcomes for the people being served.
Key Takeaways
- OpenAI's automated cybersecurity system found 948 Linux kernel vulnerabilities in 30 hours, costing $1.21 million in API credits
- AI agents now turn software patches into working exploits in hours instead of weeks, with disclosure-to-attack time dropping to one day in 2024
- AT&T's Connected Life became the first home security product to integrate with Google Home Premium via Gemini for Home, requiring a $29.99 monthly monitoring plan
- AI assistance raised customer-support productivity by 15% overall and about 30% for less experienced workers
- A two-month employee with AI help matched the performance of a six-month veteran, closing roughly three-quarters of the education-based productivity gap
- Only about 10% of workers highlight AI skills on their resumes, according to ZipRecruiter economist Nicole Bachaud
- Over 50 universities from 20+ countries endorsed a declaration at the Global University Presidents Summit 2026 in Hong Kong, emphasizing liberal arts alongside AI education
- SAP research found 69% of organizations say AI expands employee duties, but upskilling is lagging
- 49% of U.S. adults use chatbots, yet experts stress AI cannot replace the human touch in care and disability services
- Entry-level workers now need judgment, trustworthiness, and verification skills to stay valuable as AI accelerates work expectations
OpenAI AI agents discover 948 Linux security flaws in 30 hours
OpenAI built an automated cybersecurity system that found 948 Linux kernel vulnerabilities in 30 hours. The system used an upcoming OpenAI model and cost $1.21 million in API credits. It reproduced 1,724 total vulnerabilities and created 1,169 working privilege escalation exploits. Human reviewers helped patch 61 of the most serious flaws within 10 days.
OpenAI AI model creates 948 Linux root exploits in one run
OpenAI's Cyber team ran an automated pipeline in mid-October 2026 against the latest Linux kernel. The system generated 948 local privilege escalation exploits at a cost of $1,213,430, about $1,279 per exploit. Researchers found 10 new exploitation techniques never seen before. By day 10, 61 vulnerabilities were patched upstream and hundreds of reports were submitted.
AI cuts exploit development time from weeks to hours
AI agents now turn software patches into working exploits in hours instead of weeks. Data shows the time from vulnerability disclosure to attack dropped to as little as one day in 2024. Open-weight models can exploit bugs for about $1,000 each. Patching has not kept pace, with median fix times for critical vulnerabilities still around 30 to 43 days.
Gemini for Home brings AI features to AT&T security system
AT&T's Connected Life system became the first home security product to integrate with Google Home Premium. Gemini for Home adds camera descriptions, natural-language video search, and daily activity summaries. The features require AT&T's professional monitoring plan at $29.99 per month. AT&T launched Connected Life in December 2025 with Google and Abode.
AI helps novices catch up but benefits capital most
AI assistance raised customer-support productivity by 15 percent overall and about 30 percent for less experienced workers. A two-month employee matched the performance of a six-month veteran with AI help. AI closed roughly three-quarters of the education-based productivity gap. However, gains may flow mostly to capital owners, widening inequality between labor and wealth.
Wild Dolphin Project uses AI to decode dolphin communication
Denise Herzing founded the Wild Dolphin Project and is now using AI to study dolphins. Her team identifies individual dolphins by their unique whistles. AI helps them analyze audio and video footage in new ways. This research builds on decades of dolphin observation and tracking.
MerchantFloor launches AI tool for quick Shopify product listings
MerchantFloor released a Product Listing Agent that creates Shopify listings in under five minutes. Merchants upload a product photo and a few notes to get started. The agent generates product images, descriptions, SEO content, and FAQs. New users get 25 free credits, and paid packs start at $20.
Cisco Partner Summit focuses on trusted AI at scale
Partner Summit 26 brought together Cisco's global partner ecosystem for three days. The event highlighted the need to close the trust gap as agentic AI grows. Cisco stressed security, observability, and ROI as key pillars. Tools like Cisco AI Defense and the Splunk-powered Data Fabric support safe AI deployment.
Experts say job interviews will include AI-related questions
Experts say job seekers should prepare for AI-related questions in interviews. Lisa Dubler of Georgetown University says candidates should discuss how they use AI now and in future roles. ZipRecruiter economist Nicole Bachaud says only about 10% of workers highlight AI skills on resumes. Experts recommend using examples and researching how companies use AI.
Global university leaders release AI education declaration
The Global University Presidents and Leaders Summit 2026 in Hong Kong ended with a joint declaration. Leaders from over 50 universities across more than 20 countries attended. They stressed the importance of liberal arts education alongside AI education. Prof. Priscilla Leung said humans must keep a leading role in the AI era.
Little Friends Explores AI and the Human Touch in Services
An article discusses how AI is changing human services. Statistics show 49% of U.S. adults use chatbots. The author emphasizes that while AI can boost productivity, it cannot replace the human touch needed for people with disabilities. The key question is whether AI improves care for those served.
Three Skills Needed for Entry-Level Jobs in the AI Era
Early-career workers now face faster expectations and broader duties due to AI. SAP research found 69% of organizations say AI expands employee duties, but upskilling is lagging. Success now depends on judgment, trustworthiness, and verification rather than just knowing more. These skills help young professionals stay valuable as AI speeds up work.
Sources
- OpenAI’s AI Cyber Agents Find 948 Linux Security Flaws in 30 Hours
- OpenAI’s upcoming AI model built 948 working Linux root exploits in a single run
- AI turned exploits from weeks to hours
- Gemini for Home adds AI features to AT&T’s smart security system
- AI Lets Novices Catch Up While the Biggest Gains Flow to Capital
- AI’s Next Trick? Speaking Dolphin
- The 5-Minute Product Listing: MerchantFloor Launches AI Agent for Small Shopify Stores
- Partner Summit 26: Building Trusted AI at Scale, Together
- You should expect questions about AI in job interviews, experts say—here's how to prepare
- Global University Presidents Issue Consensus Declaration on Artificial Intelligence
- Little Friends: Artificial Intelligence and Human Services… a Lot To Think About!
- Three Skills That Define the New Entry-Level Job in the AI Era
Comments
Please log in to post a comment.