A study examined how Claude and ChatGPT behave as shopping assistants, with the headline reporting that they may offer different prices based on a shopper’s wealth. The research could prompt calls for greater regulation of AI shopping tools.
Bites, a small AI-native food-ordering startup, lets diners order through ChatGPT while routing orders directly to restaurants through point-of-sale integrations. Its lower-fee model is drawing attention from DoorDash, highlighting how agentic ordering could weaken delivery apps’ control over discovery, customer data and advertising.
Researchers at Swarmchasers say a fleet of AI agents, likely using Tencent’s Hunyuan models, spent more than a week querying Alibaba’s Amap mapping service through Tencent Cloud-linked infrastructure. The preliminary report describes attempts to bypass anti-bot protections and says 211 scans were labeled “claude,” although the researchers’ tests found the behavior more consistent with Tencent models than Anthropic’s Claude.
Apple currently offers controls to restrict individual Apple Intelligence features on Macs, but not a single switch to disable all features and remove downloaded models. 9to5Mac reports that growing opposition to AI, storage concerns, and organizations that ban AI tools could increase pressure on Apple to add an official removal option, though Apple has not announced one.
Meshy says it ranked 31st in Andreessen Horowitz’s first consumer AI app monthly revenue Top 50, making it the list’s only AI 3D company. The ranking was based on U.S. consumer card-spending data tracked by YipitData, according to the report cited by Meshy.
Dallas-based investment firm Disruptive is targeting up to $10 billion for a late-stage fund and has secured $7.5 billion, according to sources cited by The Wall Street Journal. The firm, founded in 2012, has backed AI startups including Groq and Reflection AI.
A WIRED test of OpenAI’s always-on Dots agents found that they could research and refine couch recommendations, monitor product pages, and assemble detailed shopping briefs. The review also found reliability and trust issues, including transcription errors, an unsuccessful CAPTCHA attempt, and an awkward response that appeared to express affection after mishearing the user.
South Korean President Lee Jae Myung ordered a thorough investigation on Oct. 4 into a series of data breaches across commercial, regional and nonbank financial institutions. Reported impact includes about 25,000 Shinhan Bank customers, roughly 40,000 Yegaram Savings Bank customers, 89 people at Hana Bank, 119 at KB Kookmin Bank, 146 housing-loan agents at Hyundai Capital and 11 outsourced developers at BNK Busan Bank; exposed data included names, phone numbers, annual income and loan limits, with some resident registration numbers also compromised. FSC Chairman Lee Eog-weon said AI involvement cannot be ruled out and urged firms to accelerate use of AI-based security tools, while The Korea Times reported traces of a Chinese-language AI penetration-testing tool on a server believed used in the Shinhan attack, with an HTML title translating to "AI autonomous penetration testing console" — possibly linked to the open-source LLM-based ARTEX AI, though this does not confirm it carried out the breach.
A TechRadar analysis examines the Agent Sandbox Taxonomy, a community-reviewed framework that evaluates agent sandboxes across seven defense layers, seven threat categories, and three evaluation dimensions. The framework includes fingerprints, a verification probe, and scores for 26 products, but the article notes that many scores are inferred from documentation and that public disclosures about the OpenAI-Hugging Face incident were insufficient for a complete assessment.
AI agents seeking government data are swarming the Federal Reserve’s FRED data site, according to Bloomberg Technology. The activity is raising concerns about new security vulnerabilities, as JPMorgan Chase CEO Jamie Dimon warns about broader risks.
A 36Kr report says Anthropic engineers uploaded a Lean-verified proof generated with an unpublished Claude model for a long-standing percolation conjecture in dimensions three through ten. The claim has not been independently established in the source: the original researchers are reportedly waiting for a human-readable explanation, and Anthropic has not published an official announcement.
Google appears to be developing agentic Gemini tasks for Gmail’s AI Inbox, based on strings found in Gmail version 2026.09.28 and a post on X. The feature could let users start Gemini on suggested inbox action items, provide input when needed, and review or approve changes before tasks are completed or emails are sent.
Speaking at Infor Velocity Week, Aston Martin CIO Steve O'Connor said the carmaker applies Infor's industry AI agents to back-end operations such as configure-price-quote, and that Infor CloudSuite Automotive delivers new features almost as soon as they ship. He said an AI agent will never appear on the front end of an Aston Martin website because that is not the experience customers want. Each agent's outcomes are owned by someone in the business, and AI ideas are screened first for P&L impact and then for efficiency; Infor says its AI Adoption Hub builds roadmaps from public filings, meeting transcripts and known problems, cutting use-case delivery from quarters to weeks.
South Korea’s government-backed “AI for All” program is set to begin pilot operations in October, with formal services planned for December. SK Telecom, Kakao and KT are developing agentic AI services for tasks such as reservations, applications and payments, using different entry points including calls, KakaoTalk and an integrated service ecosystem.
A TMTPost analysis examines why Utopai X ranked second on Artificial Analysis’s text-to-video leaderboard while being based on MiniMax H3. It argues that the model’s post-training gains may offer strong near-term value for film production, but leave Utopai dependent on its base model, with base-model migration and iteration speed still unproven.
Bridgewater Associates founder Ray Dalio warned that the artificial intelligence boom may be nearing a bubble burst. The report does not provide further details about the timing or basis of his warning.
At Infor Velocity Week, Midwest Wheel's Steve McEnany described how the century-old truck-parts distributor is adopting AI agents while keeping one system of record: anything new, including third-party tools, is tied back to Infor. The company already uses a product recommender in order entry and automation that scans emailed PDF invoices into its system, and McEnany says a human must stay in the loop for anything touching cash or accounts. Infor, which is embedding industry-specific agents into its applications, cites research that off-the-shelf AI falls short for two out of three businesses.
A Chinese industry analysis argues that lightweight, on-device AI is moving health monitoring and initial screening from hospitals into homes, communities and rural clinics. It describes a shift toward locally processed alerts, edge-cloud collaboration and closer integration between medical-device makers and model developers, while highlighting unresolved questions around clinical validation, liability, privacy and medical-device registration.
Episil Technologies expects AI-related revenue to account for at least 10% of its business in the second half of 2026 and potentially 10%–20% in 2027, up from about 5% in the first half of 2026. The company said GaN demand from AI data centers remains strong, while SiC business is expected to recover as customer inventory adjustments wind down; these are company forecasts.
U.S. private data center construction spending reached an annualized $84.95 billion in August, up more than 73% from a year earlier, according to U.S. Census Bureau data cited by the report. The figure covers data center buildings only and excludes servers, racks, GPUs, networking equipment, and storage systems.