Journalism begins where hype ends

,,

The rise of powerful AI will be either the best, or the worst thing, ever to happen to humanity. We do not yet know which."

—Stephen Hawking

‘Cognitive Offloading’ to AI Might Hamper Students’ Learning: Stanford Research

AI is reshaping education, raising questions about its impact on students’ learning, creativity and independent thinking.

The report found that excessive reliance on AI could come at the cost of students’ ability to think critically, solve problems creatively and learn independently.

Canada Ranks 5th in Commonwealth Youth AI Readiness, But Education Gaps Remain: Study

Canada ranks fifth in the Commonwealth for youth AI readiness, but gaps in education and practical AI training remain, according to the Ayre Index 2026.

Canada has ranked fifth among 56 Commonwealth countries in youth AI readiness, scoring 70.2 out of 100 in the inaugural Ayre Index. However, the study identifies significant gaps in education and training, raising questions about whether young Canadians are adequately prepared to use AI in education and the workplace.

AI Demand Drives Tech Hiring Growth, 86% Business Leaders Plan Expansion: Randstad

Technology professionals collaborate in an increasingly AI-driven workplace as businesses expand hiring and seek specialised artificial intelligence skills. (Image: Representative)

Randstad’s report found that 86% of business leaders plan to expand their technology workforce as AI adoption accelerates. Entry-level tech job postings requiring AI skills have increased more than fivefold since early 2025, while demand for AI agent skills has surged by over 1,300%.

Sarvam AI Launches Prism, a 44ms Decision Model Inspired by TypeSafe’s Jev

Sarvam AI launches Prism, a System 1 decision model built on Sarvam 105B, entering the emerging decision-model category introduced by TypeSafe's Jev. The company claims a median response time of 44 milliseconds on an NVIDIA B200 GPU.

Sarvam AI has launched Prism, an open-weight System 1 decision model built on its Sarvam 105B architecture, entering the emerging decision-model category pioneered by TypeSafe’s Jev.

MIT Researchers Show How GPT-6 Astra Can Read and Edit Neural Network Weights Without Training

MIT researchers have demonstrated Weight-Space Programming, a technique that allows frontier AI models to interpret and directly modify the numerical weights governing neural network behaviour.

MIT researchers have demonstrated how frontier AI models can directly interpret and modify neural network weights without conventional gradient-based training. In experiments, OpenAI’s GPT-6 Astra corrected reasoning loops in Qwen3.5-0.8B, increasing its MATH-500 score from 7.9% to 48.1% by modifying just 0.016% of its weights.

Drax, Brookfield Plan £20 Billion AI Data Centre at UK Power Station

Drax Power Station in North Yorkshire, UK, where Drax, Brookfield and Radiant plan to develop a data centre with more than 1GW of capacity. Image: Drax

Drax and Brookfield plan to develop a 1GW AI data centre at the UK’s Drax Power Station. The project could attract more than £20 billion in investment, including £2.5 billion for its first phase, with initial operations targeted for 2028.

OpenAI Reported Users to Police 200 Times in 3 Months; Restricts ChatGPT Information on Mass Shooters

OpenAI's latest safety measures will restrict ChatGPT responses about previous mass shootings and their perpetrators, while introducing stricter safeguards for teenage users amid concerns over AI-assisted violent attacks.

OpenAI reported more than 200 imminent threats of violence to law enforcement between July and September 2026.

Claude Now Checks Its Own Conversations for Misuse; Self-Assessment by Anthropic

Anthropic says a technique called self-assessment now has Claude review its conversations after responding, flagging potential misuse with a harm category and priority score for human review.

Anthropic says a technique called self-assessment now has Claude review its conversations after responding, flagging potential misuse with a harm category and priority score for human review. Running on a large share of Claude traffic since June, it has surfaced child abuse material and a romance-scam operation, though Anthropic admits it cannot stop determined attackers.

PM Modi Urges AI Safety, Asks Scientists to Verify AI-Generated Answers

Indian Prime Minister Narendra Modi has called for stronger AI safety measures and urged scientists to verify AI-generated answers independently.

Speaking at the Emerging Science, Technology and Innovation Conclave 2026 in New Delhi, Modi said AI should accelerate research without replacing critical thinking and scientific validation.

UK to Invest £40 Million in Security Unit to Tackle Rogue AI Agents

The UK government plans to establish a National Agentic Security Unit with £40 million in funding to address security incidents involving rogue AI agents.

The UK government plans to invest £40 million in a National Agentic Security Unit to detect, investigate and respond to rogue AI agents. The funding forms part of a £700 million-plus AI investment package that also covers AI biosecurity, cybersecurity, model testing and sovereign AI development.