Welcome to the Global Advisors A.I Voices podcast. I'm Maya.

Today is Wednesday, the eleventh of November 2026.

This is your morning catch-up on A.I. Please like and follow the show for your daily A.I update.

Nate B Jones, host of A.I News and Strategy Daily, forecasts that compute availability will become an existential threat for non-lab companies over the next 6 to 12 months, particularly if they are under attack. As cyber attackers gain advantages from A.I, boards and leaders are under immense pressure to ensure their defenders have access to the best available models and inference speed.

Jones argues that the gating of models inside labs due to cyber risks will widen the gap between internal lab capabilities and accessible models.

This dynamic makes 2027 budget planning for businesses very tricky. Jones predicts that a long-drawn-out investigation into chip sales, such as the Nvidia probe, could make it more difficult for sales to proceed over time. This would affect the ability of companies to plan for A.I precisely when they need cyber defense the most, creating a scenario where security infrastructure lags behind the sophistication of A.I-driven threats.

Nathan Benaich, author of the State of A.I Report, suggests that users may find it easier to trust established financial platforms like Revolut over nascent A.I agent startups. Speaking on The MAD Podcast, Benaich stated he would trust Revolut more than a third-party startup that has only existed for 12 months to handle purchases, arguing that the startup likely lacks proper information security.

He noted that he does not provide his data to these startups because he does not trust them and believes he can achieve similar results using ChatGPT.

Benaich argues that the probability of a startup surviving for 10 years and being held legally accountable for issues is very low compared to established companies like Meta, which has a history of paying out for issues.

This highlights a trust deficit for A.I-native consumer applications, where users are hesitant to hand over sensitive financial data to entities that lack the institutional memory and legal accountability of incumbent tech giants.

Anthropic has entered into a deal valued at approximately 3.4 billion dollars to secure 1.7 gigawatts of heavy-duty gas-fired power generators from OnePWR Solutions for its data centers. The agreement includes engineering and construction capabilities provided by the supplier to install the generators, though Anthropic did not acquire OnePWR itself.

This move is intended to bypass long grid connection queues and address power shortages that are hindering data center expansion.

Anthropic took over Caterpillar equipment orders originally placed by OnePWR two years ago, as OnePWR had needed financing to complete those transactions. Sebastian Moss, who reported on the deal, described it as the largest known single purchase of power-generation equipment by a frontier lab, intended to give Anthropic more control over its data center portfolio.

The generators are expected to be ready within the next two years, with OnePWR assisting in the installation. High demand for Anthropic's Claude A.I models has driven this urgent need for capacity. Reuters was unable to immediately verify the report, and neither Anthropic nor OnePWR provided comments.

Google Research published a peer-reviewed study in Nature regarding a smartwatch system that identifies likely hypertension using optical and motion sensors without requiring cuff calibration. In a study of 196 patients, the smartwatch achieved an 80.6 percent positive predictive value, outperforming office blood pressure checks, which achieved 77.8 percent.

The smartwatch also demonstrated higher sensitivity than office readings, while maintaining comparable specificity.

The study was trained on extensive sensor data and caught more cases passively than office readings. Google researchers stated the findings helped inform the Health Guardian Blood pressure trends feature, but explicitly noted the work is exploratory and not indicative of any existing or upcoming commercial product feature.

The Blood pressure trends feature requires Google Health app version 5.09 or higher and works on Pixel Watch 3 and later, Fitbit Edge, and Fitbit Air, but not third-party wearables.

Oura announced it is acquiring A.I-native primary care company Counsel Health for 625 million dollars in cash. The deal comes a few weeks after Oura postponed its up to 2.2 billion dollar IPO indefinitely, citing uncertainty in the IPO market. The transaction unifies Oura's continuous physiological monitoring with Counsel Health's clinically supervised medical A.I and employed, board-certified physician workforce.

The combined offering embeds on-demand clinical escalation directly within the Oura App. Members can interact with Counsel's medical A.I for triage, questions, and guidance, with the ability to escalate seamlessly to a board-certified Counsel physician within minutes. The acquisition is subject to customary adjustments, with closing scheduled before the end of calendar year 2026.

This move signals a shift from passive health tracking to active, A.I-assisted primary care delivery.

Kirt McMaster, the co-founder behind the Cyanogen project that once challenged Android, is launching a new venture called Kora Systems. The company is developing a Linux-based desktop operating system specifically designed for the age of artificial intelligence. Scheduled to launch in 2027, Kora aims to be a free-to-use alternative to Windows and macOS, targeting users who rely heavily on A.I models and agents.

Backed by investors including Samsung and Square Enix, the team announced the project. Kora Systems argues that existing operating systems like Windows and macOS cannot fully integrate A.I because of their closed architectures. The company contends that an open Linux foundation allows for the kind of system-level transformation needed to embed A.I deeply into the user experience, rather than just running it as an application on top.

New joint research from OpenAI and Upwork reveals a significant shift in the freelance labor market driven by artificial intelligence. The study found that A.I-related fixed-price freelance postings had budgets 52.4 percent higher than comparable non-A.I postings.

However, the adjusted gap in advertised hourly rates for A.I-related work was only 1.5 percent, suggesting that while the total value of these projects is higher, the hourly premium for A.I skills is currently modest.

Instead of higher hourly rates, A.I-related hourly contracts involved 40.6 percent more hours during the first 90 days after hire. The research also noted that A.I-related postings averaged more distinct task clusters and featured historically uncommon task pairs more frequently than non-A.I jobs. The authors caution that these datasets describe associations rather than causal effects and do not represent the entire labor market.

Devon Thomas founded Abliteration A.I, a company dedicated to removing safeguards from A.I models. Thomas built a 20,000 dollars computer cluster to test open-weight models and found they refused many queries, prompting him to seek ways to modify them. His work is part of a growing subculture of engineers who share methods to jailbreak popular chatbots and modify open-weight models to remove restrictions.

This trend challenges the control major A.I companies like OpenAI and Anthropic exert over their models, which often charge extra for access to versions without cybersecurity guardrails and restrict model distillation. Will Knight, writing for Wired, suggests that the push toward fewer controls may lead to a world where big A.I companies' efforts to lock down models are questioned as potentially self-serving, rather than purely safety-driven.

Amazon-owned Ring is expanding beyond doorbells with a new robotics division. The company announced two devices: a rolling enterprise security robot called the Rover, priced at 1,000 dollars, and a home security drone priced at 270 dollars. The Rover is scheduled to launch in mid-December, while the drone arrives in late January.

Ring C E O Jamie Siminoff stated that moving intelligence to the cloud resolved issues with the earlier 'Ring Always Home Cam' project. However, the company is currently seeking FCC certification for the drone, asserting that its United States-assembled product complies with restrictions.

Fireworks A.I has disclosed a second security breach, this time exposing resource metadata, billing information, and incomplete model weights and datasets for some customers. The incident began when a third-party cloud provider hosting Fireworks R and D machines was breached, leading to unauthorized access on October 24. Fireworks stated that customer payment card numbers were not accessed because the company does not store them.

The company detected the activity on the same day and implemented remediation measures, reporting no further evidence of unauthorized access as of October 25. This follows an earlier incident in early October where attackers accessed environment variables containing API keys and tokens, linked to financially motivated 'LLMJacking' threats.

That's A.I Voices for today. I'm Maya. Thanks for listening.

Follow the show on Spotify, YouTube, Apple Podcasts, Amazon Music, or your favourite podcast platform, and explore the original voices and sources in the episode notes. See you tomorrow.