Federal Reserve Chair Kevin Warsh says AI could reshape jobs and markets in several ways, but stresses that these are scenarios, not forecasts. Productivity gains could spread widely while employment stays near full capacity; growth could instead come with worker displacement and higher unemployment; or disappointing results after heavy investment and stock-market gains could pull valuations down and slow the economy. Warsh says the Fed will neither promote nor oppose AI. A San Francisco conference on Nov. 16, 2026, brought outside experts into the Fed’s assessment, with five working groups expected to report near year-end. Speakers differed on adoption speed: Marc Andreessen pointed to health-care and education regulation as limits on replacing workers, while former Commerce Secretary Gina Raimondo criticized state efforts to slow data-center construction. Warsh offered no immediate conclusions about interest rates.
AI investment is also changing workforce needs. An Experis report says employers are seeking automation architects, data ethicists, IT support staff and more cybersecurity specialists as they adopt AI tools and strengthen defenses against cyberattacks. In finance, Hong Kong Securities and Futures Commission chief Julia Leung Fung-yee warned that agentic AI could bring market volatility and cybersecurity risks, and urged regulators and industry to collaborate.
Several developments focus on how AI is built and secured. NVIDIA has released NVKE, which manages Kubernetes clusters on GPU infrastructure, and NVKR, which builds the operating-system and container images and packages they require. Both are Apache 2.0-licensed public previews within NVIDIA’s DSX AI Factory Platform, not production-ready; they were tested with Kubernetes 1.36. Soracom, meanwhile, offers security assessments for IoT and physical-AI systems such as robots and drones. AI agents conduct penetration tests, with engineers setting the scope and reviewing results; testing requires written customer authorization.
Researchers are applying AI and machine learning to improve microbial production of useful nanomaterials, a potentially less hazardous and energy-intensive alternative to some chemical methods. Scaling remains difficult because biological processes are complex and production can be inconsistent. Security researcher Peter Stöckli says overly prescriptive prompts can limit AI security audits. His team maps software components, entry points and user actions, checks findings with researchers and proof-of-concept evidence, and reports a 95% fix rate; GitHub is developing tools to help maintainers manage reports.
Governments and universities are also expanding AI-related education and oversight. Connecticut lawmakers are considering giving applicants rejected after AI screening access to the information used and a chance to correct errors. Experts also call for clear responsibility among employers and developers and meaningful human review. Himachal Pradesh University is adding AI courses and research: a Rs 5 crore grant announced in July 2026 supports a BTech AI course, a September agreement with NIIT Foundation provides training, and the university has approved PhD programmes at its Centre for AI and Cyber Physical Systems.
In India, Sarvam has opened Token Factory, a managed inference platform serving eight open models from GPU clusters in the country through an OpenAI-compatible API. Sarvam says DeepSeek V4.1 Flash has served more than 450 billion tokens a day, with 95.7% of prompt tokens coming from cache. It reports 330 tokens per second for Sarvam 105B with one user and 6,177 tokens per second per GPU for Kimi K3.1 Pro, versus 3,213 for InferenceX; the company says these benchmarks are self-reported. Introductory Sarvam 105B prices are ₹10 per million input tokens and ₹25 per million output tokens.
Key Takeaways
- Kevin Warsh outlined three possible AI effects: broad productivity gains, growth paired with worker displacement, or an economic slowdown if returns disappoint; he says these are scenarios, not forecasts.
- The Fed will neither promote nor oppose AI. Five outside-led working groups are expected to report near year-end after a Nov. 16, 2026, San Francisco conference.
- Experis reports demand for automation architects, data ethicists, IT support workers and cybersecurity professionals as companies invest in AI.
- Hong Kong regulator Julia Leung warned that agentic AI could increase market volatility and cybersecurity risks, urging cooperation between regulators and industry.
- NVIDIA’s Apache 2.0-licensed NVKE and NVKR projects support Kubernetes on GPU systems; both are public previews and not production-ready.
- Soracom launched security assessments for IoT and physical-AI systems, using AI agents for penetration tests and engineer review; customer authorization is required.
- Connecticut lawmakers are considering letting applicants rejected by AI hiring screens review and correct the information used, with experts calling for human oversight.
- Himachal Pradesh University is expanding AI education, backed by a Rs 5 crore grant for a BTech AI course and new training and research programmes.
- Sarvam Token Factory serves eight open models through an OpenAI-compatible API from Indian GPU clusters; Sarvam reports DeepSeek V4.1 Flash throughput above 450 billion tokens daily, with 95.7% of prompt tokens cached.
- Researchers are using AI to improve microbial nanomaterial production, while security researcher Peter Stöckli warns that overly detailed audit prompts can constrain findings; his team reports a 95% fix rate.
Warsh outlines three ways AI could reshape jobs and markets
Federal Reserve Chair Kevin Warsh described three possible outcomes as artificial intelligence develops, and said they were scenarios, not forecasts. In one, AI raises productivity and spreads gains widely while the United States stays near full employment. In another, AI boosts growth but displaces workers and pushes unemployment above levels consistent with full employment. If AI disappoints after major investment and stock market gains, falling valuations could slow the economy and raise unemployment. Warsh said the Fed would neither promote nor oppose AI, and its five outside-led working groups are expected to report near year-end for the Federal Open Market Committee to consider.
Fed Chair Warsh weighs AI’s risks to jobs and the economy
Federal Reserve Chair Kevin Warsh said the Fed must prepare for AI’s effects without promoting or opposing the technology. He described possible outcomes ranging from broad economic gains to major job losses and slower productivity if AI disappoints. Warsh said AI could speed up innovation, but higher unemployment and harmful uses of AI could cause serious economic disruption. He called the technology “super intelligence,” using language preferred by Donald Trump and Elon Musk. Other central bankers, including Fed Governor Michael Barr, also discussed AI’s effects on Monday.
Warsh turns to outside experts to assess AI’s impact on the economy
Federal Reserve Chair Kevin Warsh convened a conference in San Francisco on November 16, 2026, to examine how AI could affect productivity, jobs and the economy. He raised the possibility that AI could displace workers and lower service prices, while speakers disagreed on how quickly businesses will adopt it. Venture capitalist Marc Andreessen said regulation in fields such as health care and education limits how much technology can replace workers. Former Commerce Secretary Gina Raimondo called state measures to slow data center construction troubling and said stopping AI would be bad for America. The conference was part of Warsh’s effort to bring outside experts into a task force whose findings could inform future Fed policy, though he offered no immediate conclusions about interest rates.
AI investment drives demand for tech and cybersecurity workers
An Experis report found that companies are hiring automation architects, data ethicists and IT support workers as investment in AI grows. Firms are also seeking more cybersecurity experts and strengthening defenses against cyberattacks. Companies are investing in AI tools and services to improve operations and compete. The report says demand for skilled technology workers and cybersecurity professionals is expected to keep growing as AI advances.
Hong Kong regulator warns of risks as agentic AI enters finance
Julia Leung Fung-yee, chief executive of Hong Kong’s Securities and Futures Commission, discussed the challenges of agentic AI at the Bloomberg Global Regulatory Forum on Monday. She pointed to risks including market volatility and cybersecurity threats. Leung said regulators and industry peers should work together to address these challenges rather than compete over who is best.
AI Helps Improve Microbial Production of Useful Nanomaterials
Microorganisms can produce nanomaterials using enzymes and other natural substances, offering an alternative to methods that may use hazardous chemicals, toxic solvents, and large amounts of energy. These bio-based materials can have better biocompatibility and surface functions than chemically made materials. But complex biological processes and inconsistent production have made large-scale manufacturing difficult. Researchers are using artificial intelligence, machine learning, and data tools to improve each stage of production.
Too Many AI Instructions Can Limit Security Audits
Security researcher Peter Stöckli said AI security audits can miss opportunities when prompts include too many instructions. His team first focused on authentication and authorization flaws, including missing access checks and insecure direct object references. The audit process maps software components, entry points, and user actions before examining code. Security researchers check every finding and report, provide proof of concept, and filter out false or lower-severity findings; Stöckli said the team has a 95% fix rate. He also said GitHub is developing tools to help maintainers handle incoming reports.
NVIDIA Releases Open Source Tools for Kubernetes on GPU Systems
NVIDIA released two open-source projects for managed Kubernetes on GPU infrastructure: NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR). NVKE manages Kubernetes clusters, while NVKR builds the operating system images, container images, and packages those clusters need. Both projects are part of the NVIDIA DSX AI Factory Platform, use the Apache 2.0 license, and are in public preview, not production ready. They were tested with Kubernetes 1.36, and NVKE uses TPM-attested node identity. A local evaluation needs 6 CPUs, 12 GiB of RAM, 100 GiB of disk, and no GPU.
Soracom Offers Security Testing and Hardening for IoT and Physical AI
Soracom launched a security assessment and hardening service for IoT and physical AI systems, including robots and drones. The service checks devices, communication paths, and servers or cloud services for security weaknesses, then helps design, apply, and retest fixes. AI agents conduct penetration tests, while Soracom engineers set the test scope and review findings before sending customers a technical report. Testing starts only after customers provide written authorization and agree to rules for the work. Countermeasures can use Soracom Virtual Private Gateway, Soracom Krypton, and Soracom Arc, and may include software or middleware changes.
Connecticut Considers Letting Rejected Applicants Review AI Screening
Connecticut lawmakers are considering expanding a law that requires employers to tell job applicants when AI screens their applications. State Rep. Roland Lemar and Sen. James Maroney said rejected applicants should be able to review the information AI used and possibly correct errors, such as a missed education credential or a resignation wrongly recorded as a firing. The state’s AI working group heard ideas from experts as lawmakers prepare for the legislative session starting Jan. 6. Experts also called for clear responsibilities for AI developers and employers, along with meaningful human review of hiring decisions.
Himachal Pradesh University expands AI courses and research
Himachal Pradesh University is adding AI to its courses and daily operations to prepare students for future jobs. In July 2026, Chief Minister Sukhvinder Singh Sukhu announced a Rs 5 crore grant for a BTech course in AI. The university signed an MoU with the NIIT Foundation in September, offering training in Data Analytics, IT Essentials and AI Digital Productivity 2.0, and approved PhD programmes at its Centre for AI and Cyber Physical Systems, along with certificate courses in AI and cyber security. Vice-Chancellor Professor Mahavir Singh said HPU already offers courses in AI and Data Science, and that five research centres established last year include an AI component.
Sarvam opens India-hosted AI inference platform for eight open models
Sarvam has opened Sarvam Token Factory, a managed inference platform that serves its own and open-weight models from GPU clusters in India through one OpenAI-compatible API. Its stack uses tools including vLLM, SGLang, Dynamo and Mooncake, with a shared cache and separate GPU pools for prompt processing and response generation. Sarvam says caching helped DeepSeek V4.1 Flash serve more than 450 billion tokens per day, with 95.7% of prompt tokens coming from cache. The company reports benchmark results including 330 tokens per second for Sarvam 105B with one user, and 6,177 tokens per second per GPU for Kimi K3.1 Pro compared with 3,213 for InferenceX. Sarvam lists introductory prices for Sarvam 105B at ₹10 per 1 million input tokens and ₹25 per 1 million output tokens, and says its benchmark figures are self-reported.
Sources
- Fed Chair Kevin Warsh warned that AI could trigger job losses or a stock bust if it disappoints
- The Fed seems to be getting worried about AI
- Analysis: The Fed's chairman sees an AI revolution coming. Now he has to figure out what it means
- Firms bolster cybersecurity, create tech jobs as AI booms
- As agentic AI seeps into finance, concerns grow over new risks and blind spots
- Artificial Intelligence Accelerates Microbial Biosynthesis of Functional Bio-Based Nanomaterials
- Too many instructions can hold back AI security audits
- NVIDIA Open-Sources NVKE and NVKR for Managed Kubernetes on GPU Infrastructure
- Soracom Launches Security Assessment and Hardening Service for IoT and Physical AI Systems
- Connecticut May Show Rejected Job Applicants How AI Screened Them
- Himachal varsity integrates AI into courses to prepare students for future
- Sarvam Launches Sarvam Token Factory: Bringing India-Hosted Inference to 8 Open Models
Comments
Please log in to post a comment.