At COP31 in Antalya, Türkiye, a global initiative co-chaired by the UN, UNESCO and Brazil launched Priorities for Action on AI and Climate Information Integrity, urging AI firms to assess and disclose their systems' impact on climate information and their energy, carbon, water and land footprints, and states to embed information integrity in national AI strategies.
A study by Inquiara Lab and India's Vipassana Research Institute surveyed 251 people entering 10-day meditation retreats and re-measured 48 a week later. It aimed to understand cognitive sovereignty—whether people can pause and decide for themselves rather than let impulse or AI decide. The sample is not representative but sheds light on preserving mental autonomy.
Startup Mansa uses AI microdramas as proof of concept to show Hollywood what Black audiences want, though critics say the studio is selling out.
Incoming Florida House Speaker Sam Garrison created an Advanced Technology & Digital Security Committee covering AI, cybersecurity and data governance, aiming to prevent costly tech failures. The House earlier rejected Governor DeSantis's AI consumer-protection proposal, arguing the federal government should act.
Callosum co-founder Dan Akarca says Nvidia will stay dominant near term but its share is likely to decline as specialized AI chips emerge. He says Callosum's software routes AI tasks to the best-suited processors, cutting costs, improving performance and reducing energy use.
A Hacker News discussion argues that the spread of AI-generated content is undermining the trust between writers and readers, focusing on how readers can no longer tell whether text is human-written and how transparency and attribution might rebuild it.
Major UK retailers are boosting hiring for AI and tech roles while rising costs force cuts elsewhere, including in-store staff. Retail accounted for almost half of tech job vacancies at consumer-facing firms this year, per Vacancysoft, more than luxury and travel.
A PhD researcher who published at A* conferences describes a poor experience with SnorkelAI's paid survey: pushy follow-ups, a muddled website, and onerous payment paperwork needing an in-person tax office visit. He advises peers to avoid it.
Hacker News users reported Claude Code was unusable, with Claude's status page later confirming a major outage that repeatedly logged users out and stalled on authentication. Some users reported service returning, and discussion focused on whether traffic would shift to rival models.
South Korean startup DeepX plans to spend more than $100 million developing its next-gen AI chip DX-M2 on Samsung's 2nm process. Design is complete and sent to Samsung Foundry; first silicon is expected by March 2027, with customer evaluation and mass production in 2028.
The author, while skeptical of AI, accepts that avoiding it is unrealistic, noting Microsoft and Apple use coding agents for OS development. The argument: AI lowers the cost of producing an app but not the work needed to make a great one, which still demands reliability and craft.
Gymnastics is leading AI-assisted judging, and officials say they would be "highly surprised" if it were not used at LA28. At the recent World Championships, however, AI difficulty detection downgraded Harry Hepworth's and Carlos Yulo's vaults from 5.6 to 5.2, prompting athletes to criticize the calls and say they are "not robots."
Société Générale CIB chief US equity strategist Manish Kabra said he will favor companies using AI over those building it in 2027, citing the historically volatile semiconductor cycle. He said corporate conviction to adopt AI is feeding into hyperscalers and frontier labs, which in turn drives the AI capex cycle.
Anna Ashton, founder of Ashton Analytics, said the Nvidia controversy will likely lead to stricter US export controls toward China and deeper restrictions on technology products.
EU leaders held their first-ever AI summit to discuss and address the challenges posed by artificial intelligence. Specific agenda items and outcomes have not yet been disclosed.
Server maker xFusion received approval from Shenzhen regulators for a $1.2 billion listing. The move shows Chinese AI compute infrastructure firms accelerating into public markets.
An Isaacus engineer released an open-source legal conflict detector that uses only encoder models and natural language inference to compare clauses across legal documents. The author argues generative models are too costly for pairwise review—two 133-clause documents would need 17,689 API calls—and suffer from weak recall and hallucination. The pipeline ran in under 20 seconds on the longest document pairs tested.
Democratic group Tomorrow Together announced a research initiative in Denver, initially funded at $1.1 million, to shape how Democrats address voters' AI concerns ahead of the 2028 presidential race. It spent $6.5 million supporting pro-regulation candidates in the midterms, contrasting with Trump's opposition to regulating AI.
A study models how artificial intelligence can trigger an irreversible breakdown in institutional fact-checking. The work signals that AI-generated content may systematically undermine newsrooms' and institutions' ability to verify information; further details were not disclosed.
Growing numbers of people, especially the young, are turning to chatbots like Gemini, Claude and ChatGPT for emotional support. Experts stress AI cannot replace human psychological care, though around-the-clock availability and non-judgmental responses drive adoption; in Germany therapy waits average 20 weeks, highlighting care gaps.
Alpha Chicago, a new Alpha Schools campus in Lakeshore East, uses an AI program called "2 Hour Learning" for two hours each morning to teach core subjects while students spend the rest of the day on life skills. Alpha Schools operates 50 campuses with 1,500 students nationwide; the Chicago site enrolls about 50 students (capacity 500) and charges nearly $55,000 tuition. About one-third of guides have education training.
The National Animal Nutrition Program is developing the Feed Value Loss Risk Assessment Tool, which combines published research and expert knowledge with a Random Forest model to estimate relative risk of feed value loss across silage, hay, and grains. The interactive decision-support app will help producers identify high-risk management practices to reduce feed losses and improve profitability.
A LocalLLaMA post says A100 prices have dropped another 20-30% since its earlier collapse warning, while RTX 6000 fell to $14,800 from an $18,000 peak. The author argues Rubin and Blackwell are 3-4x more efficient per watt with more financing available, A100s cannot be used for inference, and remaining buyers are price-sensitive hobbyists and small renters.
A multimodal AI accurately detects Alzheimer's disease but stumbles on staging, indicating potential for diagnosis but limitations in assessing disease progression.
Malaysia is investigating whether AI chips imported by a unit of Singapore-based Megaspeed International were later reexported, potentially implicating export control compliance.
Claude Code 2.1.323 removed the subagentStatusLine setting with no replacement or migration path. It was the only way users could see each subagent's live context use, token and dollar spend, status and prompt-cache state. The configured command silently stops running, and users discover it only when the agent panel goes blank.
Singtel Group CFO Arthur Lang discusses growth strategy after 1H earnings and potential business impact from Nvidia leadership changes.
Brave Search API now supports pay-per-request agentic payments via the open x402 protocol, letting AI agents pay in USDC on Base without API keys or accounts. Endpoints including LLM Context cost $0.005 per request ($5 per 1,000).
Reuters reports that Trump's AI task force met on Thursday to discuss AI-powered cyberattacks. The report gave no details on conclusions or follow-up measures.
Singapore charged a fifth person in an AI server fraud case involving exports of servers that may contain Nvidia chips to Malaysia, the Straits Times reported. The case highlights enforcement of AI chip export controls.
Load more