Read the full analysis: The Latest Developments In AI And Work: November 14, 2026 on ThorstenMeyerAI.com
TL;DR
A November 14 roundup from ThorstenMeyerAI.com links several developments affecting work: AI risk concerns among financial market contacts, competing approaches to robot control, and a breach exposing Uber drivers’ license information through a vendor. The FAA’s nationwide ground stop is also covered, but the source says officials found no evidence it was AI-related and the precise cause remains unresolved.
ThorstenMeyerAI.com’s November 14 roundup reports that AI is appearing in financial risk discussions and workplace vendor chains, while a breach at an Uber AI vendor exposed drivers’ license information. The roundup also contrasts general-purpose and specialist robot models and reports that the FAA’s nationwide flight halt stemmed from telecom maintenance, with officials saying there was no evidence of an AI connection.
The Federal Reserve’s latest Financial Stability Report said 87% of 23 market contacts surveyed named AI as a risk, compared with 50% in the spring. The report presents those responses as the contacts’ views, not as the Fed’s own assessment. Separately, the Fed reported that debt issuance tied to advanced computing had passed $680 billion by October 2026. Australia’s APRA regulator said its 2027 stress test will examine AI-driven disruption; it also found that few institutions had tested how to switch away from a critical AI provider.
In robotics, Fortune reported that OpenAI had broken up its standalone robotics team, saying its general models, including GPT-6 Astra, could be applied directly to robot control. Two days earlier, Runway released Praxis-1, a specialist robot model with open weights. Runway’s own test reported that GPT-6 Astra took 15.6 times longer and cost 51 times more per completed task than Praxis-1. The source notes that these are vendor-reported results. Artificial Analysis separately measured GPT-6 Astra taking an average of 30 minutes per task on a real robot arm, at $16.36 per task; a person took about 20 seconds in that test.
Uber has begun notifying US drivers whose data was taken in an October breach that began at vendor Fireworks AI, according to Uber’s account to TechCrunch. Breach notices reviewed by TechCrunch said driver’s license information was stolen. Notices from three states named 7,870 affected people, a count the source describes as a minimum rather than the total. The roundup also covers the FAA’s November 13 ground stop, which lasted about four hours and 23 minutes across US airports. Transportation Secretary Sean Duffy attributed its start to contractor L3Harris’s routine work switching lines from copper to fiber at a remote FAA site, according to news reports; the specific equipment or installation failure had not been determined.
AI Risk Reaches Workers and Vendors
The developments matter because AI’s effects on work are not limited to whether a model can perform a task. Financial exposure, third-party providers and operational resilience also shape how firms adopt the technology and who bears the consequences when systems fail. APRA’s planned stress test puts disruption from AI adoption into financial institutions’ planning, while its finding about provider switching points to a practical dependency: firms need people and procedures capable of moving essential work elsewhere.
The Uber breach illustrates a different kind of exposure. Drivers may have little visibility into the vendors a platform uses, yet their personal information can be held or processed across that chain. The reported theft of license details is consequential because such identifiers cannot be replaced as readily as passwords. The roundup’s interpretation is that responsibility for protecting worker data extends through vendors; the source does not establish what legal or contractual duties apply in this case.
Robot performance comparisons also have labor implications, but the available figures are tests, not proof of broad workplace readiness. A model that is slower or more expensive in one evaluation may still have other uses, and vendor benchmarks need independent comparison. The source argues that deployment experience and real-world data held by robotics companies and workplaces could shape which systems prove useful. That is an interpretation, not a measured forecast of job losses or gains.
From Stress Tests to Robot Trials
The roundup is a collection of developments published on November 14, rather than a single policy announcement or research paper. Its financial figures come from the Fed’s report and APRA’s stated plans. The labor-economics framing—that AI risk now sits in balance sheets, supplier networks and the work of maintaining existing systems—is the publisher’s analysis, which it distinguishes from reported facts.
For the robotics comparison, Runway released Praxis-1 two days before the roundup, while the OpenAI team change was reported by Fortune. The speed and cost comparison comes from Runway’s own test, and the real-arm result comes from Artificial Analysis’s AA-Robotics v0.1 evaluation dated October 14, 2026. These measurements use different tests and should not be treated as a direct, complete measure of performance across workplaces.
The FAA outage provides a separate infrastructure example. The source says the FAA tracker showed 68% of telecom connections replaced as of October 15, 2026—3,508 of 5,170. The flight halt occurred during a wider modernization effort, but that timing alone does not establish that the replacement program caused the outage. Officials’ reported account connected the incident to a contractor’s routine line-switching work.
“zero evidence of anything AI-related”
— FAA Administrator Bryan Bedford, as quoted by CNN
Key Figures and Causes Still Open
Several parts of the roundup remain unsettled. APRA has not published the figures or assumptions it will use in its 2027 AI disruption scenarios. The Fed’s 87% figure reflects 23 market contacts’ views; it is not a survey of all financial institutions or the Fed’s own estimate of the likelihood of a crisis.
The full number of Uber drivers affected is not known from the state notices counted in the source. The 7,870 figure is a floor based on notices from three states, not a confirmed total. The roundup does not provide a complete account of what information beyond license details may have been accessed or what investigation steps have been completed.
For the FAA outage, reports had not established whether a router was faulty or installed incorrectly. The available account identifies work switching copper lines to fiber but does not settle the precise technical failure. Robot-model results also remain limited to particular tests: Runway supplied the comparison with Praxis-1, and neither cited evaluation establishes how either model performs across routine jobs in varied real-world settings.
Stress Tests and Incident Follow-Up
APRA’s planned 2027 stress test is the clearest scheduled next step. Its published scenarios and assumptions should show how the regulator will examine disruption and whether institutions can manage dependence on critical AI suppliers. The source does not give a release date for those details.
Uber’s notification process and further breach disclosures may clarify the total number of affected drivers and what information was involved. For the FAA, a fuller technical account would be needed to determine what failed during the line switch and whether any corrective work followed. Further independent robot evaluations could help establish whether the reported speed and cost differences hold beyond the tests cited in the roundup.
Key Questions
Did the Fed declare AI a financial stability threat?
No. The report said 87% of 23 market contacts named AI as a risk. The source explicitly distinguishes those respondents’ views from the Fed’s own assessment.
How many Uber drivers were affected by the vendor breach?
The source counted notices covering 7,870 people in three states, but describes that number as a floor, not the confirmed total. TechCrunch reviewed notices saying driver’s license information was stolen.
Was the FAA ground stop caused by AI?
The reported account linked the outage to contractor L3Harris’s work switching lines from copper to fiber. FAA Administrator Bryan Bedford said there was “zero evidence of anything AI-related,” according to CNN. The precise technical failure remained undetermined.
What did the robot-model tests find?
Runway reported that GPT-6 Astra took 15.6 times longer and cost 51 times more per completed task than Praxis-1 in its test. Those are vendor-reported results; a separate Artificial Analysis test measured GPT-6 Astra averaging 30 minutes per task on a real robot arm.
Source: ThorstenMeyerAI.com