AI governance and regulation are seeing significant developments across the globe. ISO/IEC 42001, published in December 2023, has become the first international standard for managing AI responsibly within organizations. Rather than testing individual AI models, the standard evaluates whether companies have proper processes for deciding how AI is built, used, and monitored. It includes 38 controls across nine areas covering AI policies, data, and system lifecycle, with certification requiring external audits and annual reviews.
On the regulatory front, Delaware is tightening rules around AI profiling in the workplace. Governor Matt Meyer signed amendments to the Delaware Personal Data Privacy Act on September 2, 2026, limiting AI profiling of job applicants, employees, and contractors. The rules take effect January 1, 2027, with civil penalties reaching up to $10,000 per violation.
Meanwhile, Ireland is positioning itself as a hub for AI investment. IDA Ireland CEO Michael Lohan highlighted the country as a top location for AI operations, referencing OpenAI's expansion that added 200 jobs and a new Applied AI Lab. He also pointed to the Scaling Physical Intelligence forum with Analog Devices and the International AI Summit in Dublin, stressing the need for collaboration and responsible AI governance.
In research and development, several notable advances are taking shape. A study across GPT, Gemma, and Qwen models found that using a separate answer call helps AI follow current evidence far better than continuing its own reasoning, improving performance by 29.9 to 60.6 points on HotpotQA and MuSiQue datasets. Separately, Linus Torvalds submitted his first LLM-assisted patch to the Linux kernel, addressing a long-standing cache contention problem that multiple developers had tried to solve before.
AI is also making strides in specialized domains. Researchers are using AI to predict climate tipping points, aiming to improve forecasting accuracy so communities can prepare for severe environmental impacts. In manufacturing, AI is adding intelligence to ERP, MES, and CMMS systems, helping companies move from systems of record toward systems of intelligence that anticipate problems and reduce disruptions. However, Geoffrey Hinton's 2016 prediction that deep learning would surpass radiologists within five years has not yet come to pass, as AI continues to outpace clinical trials, especially in radiology, prompting experts to call for faster adaptation and better integration of AI in medical research.
Key Takeaways
- ISO/IEC 42001 is the first international standard for responsible AI management, published in December 2023, with 38 controls across nine areas and voluntary certification through external audits.
- Delaware signed amendments to its Personal Data Privacy Act on September 2, 2026, limiting AI profiling of workers with penalties up to $10,000 per violation, effective January 1, 2027.
- IDA Ireland CEO Michael Lohan promoted Ireland as a top AI investment location, citing OpenAI adding 200 jobs and opening a new Applied AI Lab.
- A multi-model study found that separate answer calls improved AI evidence following by 29.9 to 60.6 points on HotpotQA and MuSiQue datasets compared to continuing self-reasoning.
- Linus Torvalds submitted his first LLM-assisted patch to the Linux kernel, solving a long-standing cache contention issue.
- AI is being used to forecast climate tipping points, with researchers aiming to improve accuracy so communities can prepare for severe environmental shifts earlier.
- AI is transforming manufacturing software by adding intelligence to ERP, MES, and CMMS systems, shifting organizations from systems of record to systems of intelligence.
- Geoffrey Hinton predicted in 2016 that deep learning would surpass radiologists within five years, but AI still has not replaced specialists, and experts urge faster clinical trial adaptation.
- OpenAI is expanding its presence in Ireland with 200 new jobs and a new Applied AI Lab.
- Experts call for more research, better training, and stronger integration of AI in medical trials to keep pace with rapid AI advancement in fields like radiology.
ISO 42001: The first global standard for responsible AI
ISO/IEC 42001 is the first international standard for managing AI responsibly within organizations. Published in December 2023, it checks whether companies have processes to decide how AI is built, used, and monitored rather than testing individual AI models. The standard includes mandatory management system clauses and a menu of 38 controls across nine areas such as AI policies, data, and system lifecycle. Certification involves external audits and annual reviews, and while voluntary, it helps demonstrate responsible AI governance.
Separate answer calls improve AI evidence following
A study across GPT, Gemma, and Qwen models found that using a separate answer call helps AI follow current evidence better than continuing its own reasoning. On HotpotQA and MuSiQue datasets, the separate call improved evidence following by 29.9 to 60.6 points. Removing reasoning traces lowered accuracy similarly in both setups, but continuation often repeated outdated conclusions. The findings show that benefiting from a trace and deferring to it are different properties that evaluations should report together.
Webinar explores expanding AI automation in clinical practice
A upcoming webinar will help imaging leaders learn how to expand AI automation in clinical settings. The panel will discuss benefits, challenges, and strategies for integrating AI into existing workflows with real-world examples. Attendees can join a live Q&A session with the experts. The event aims to support organizations moving beyond isolated AI tools toward streamlined clinical workflows.
Linus Torvalds submits first AI-assisted patch to Linux kernel
Linus Torvalds posted his first LLM-assisted patch to the Linux kernel development branch. The patch addresses a long-standing issue with massive cache contention when many CPUs read the same file. Several developers had attempted to solve this problem before with varied results. Torvalds used AI tooling to help draft the solution after previously issuing pro-AI directives for the project.
AI raises the standard for manufacturing technology
AI is transforming manufacturing software by helping companies turn operational data into action faster. Rather than replacing ERP, MES, or CMMS systems, AI adds intelligence to these platforms to recognize patterns and support better decisions. The technology evolution moved from systems of record to systems of visibility and now toward systems of intelligence and orchestration. This shift helps manufacturers anticipate problems, coordinate work across facilities, and reduce operational disruptions.
AI climate simulation forecasts tipping points
Researchers are using AI to predict climate tipping points. These are critical thresholds where small changes lead to large environmental shifts. The method aims to improve forecasting accuracy. It could help communities prepare for severe climate impacts earlier.
Delaware tightens AI and profiling rules for employers
Governor Matt Meyer signed amendments to the Delaware Personal Data Privacy Act on September 2, 2026. The new rules limit AI profiling of job applicants, employees, and contractors. They take effect on January 1, 2027. Violations can lead to civil penalties up to $10,000 per violation.
AI outpaces clinical trials, experts urge faster adaptation
AI is advancing faster than clinical trials can keep up, especially in fields like radiology. Geoffrey Hinton predicted in 2016 that deep learning would surpass radiologists within five years. While AI helps improve outcomes, it has not yet replaced specialists. Experts call for more research, better training, and stronger integration of AI in medical trials.
IDA Ireland CEO promotes AI investment and responsible innovation
IDA Ireland CEO Michael Lohan highlighted Ireland as a top location for AI investment. He referenced OpenAI adding 200 jobs and opening a new Applied AI Lab. He also noted the Scaling Physical Intelligence forum with Analog Devices and the International AI Summit in Dublin. Lohan stressed the need for collaboration and responsible AI governance.
Sources
- ISO 42001: The First AI Standard
- When Reasoning Traces Conflict with Evidence: How Answer Configuration Shapes Evidence Following
- Webinar: Imaging leaders explore how to expand AI automation in clinical practice
- Linus Torvalds posts his first AI-assisted patch to the Linux kernel, synthetic test clocks 23x performance leap — the bossman makes use of AI tooling himself after issuing clear pro-AI directives
- How AI Is Raising the Standard for Manufacturing Technology
- Forecasting 'tipping points' with AI climate simulation
- Delaware Employers Face Tighter AI and Profiling Rules Effective January 2027
- AI is Moving Faster Than Clinical Trials. Here’s How We Can Catch Up.
- IDA Ireland CEO highlights AI investment and responsible innovation opportunities
Comments
Please log in to post a comment.