AI news for Friday, 20 November 2026
The day’s AI news in one list, the biggest stories first. Each story links to its short read and to the outlets that reported it.
- 1 · Policy & DramaTrump threatens tax on AI data centers built outside US
President Trump announced he would impose a "tremendous tax" on companies that build artificial intelligence data centers outside the United States, citing concerns that allowing this would harm the country's competitiveness.
Reported by The Verge
- 2 · BenchmarksAI claimed a solution to the 167 year-old Riemann Hypothesis, and a human did, too
A mathematician working at an AI company and a human mathematician at a different institution have both announced solutions to the Riemann Hypothesis, a mathematical problem dating back to 1859 that concerns hidden structure in prime…
Reported by TechCrunch and 1 more
- 3 · Policy & DramaResearcher plants 2m canaries into the AI supply chain. A privacy disaster surfaced.
A researcher conducted an experiment by embedding unique trackable secrets into millions of AI prompts sent through networks of AI services to test whether sensitive information stays private.
Reported by Ars Technica
- 4 · Policy & DramaMeta reconsiders 2030 net-zero goal amid record AI emissions
Meta is reconsidering its original commitment to achieve net-zero emissions by 2030 due to record increases in greenhouse gas emissions from its AI data center expansion.
Reported by Ars Technica
- 5 · BenchmarksHebbia sees improvements in financial research with GPT-6 Astra
A financial research startup tested a new AI model to improve how bankers analyze acquisition deals by finding relevant source documents and creating presentation slides.
Reported by OpenAI
- AI deals compress to 30 days as data infrastructure becomes critical
Data infrastructure has become the critical bottleneck in AI deployment, shifting from a back-office purchase to a central driver of how quickly GPU investments pay off.
Reported by SiliconANGLE
- 7 · Policy & DramaOpenAI strengthens defenses against adversarial distillation
Adversarial distillation occurs when bad actors use a frontier AI model's outputs without authorization to train another model with similar capabilities, then deploy it without comparable safeguards.
Reported by OpenAI
- 8 · BenchmarksMedChemBench tests AI for drug candidate identification
A new benchmark called MedChemBench tests whether AI models can help scientists identify promising drug candidates by evaluating how well they read molecular structures, interpret experimental results, and select which compounds to test…
Reported by OpenAI
- Agentic AI drives enterprises to own computing hardware
Agentic AI workloads are creating renewed demand for enterprises to own and operate their own computing hardware rather than renting cloud services.
Reported by SiliconANGLE
- 10 · Models & ReleasesRadixArk Ships DaisyRose-27B, a 27B Agentic Coding Model Trained With Reinforcement Learning Only
RadixArk released DaisyRose-27B, a coding model with 27.8 billion parameters that was trained using reinforcement learning only, without other common training methods.
Reported by MarkTechPost
- 11 · Funding & DealsAnthropic goes public, but is it too late to save the AI trade?
An artificial intelligence model maker completed its initial public offering, with shares rising 17 percent on the first day of trading.
Reported by SiliconANGLE
- 12 · Models & ReleasesNVIDIA releases Nemotron-Labs-Sigma diffusion model
A continuous diffusion language model is a type of AI system that generates text by working in a smooth, learned embedding space rather than jumping between discrete tokens like most other diffusion models do.
Reported by MarkTechPost
- 13 · RoboticsRobotics enters its post-Jensen Huang era
NVIDIA's co-founder Jensen Huang resigned from the company in November, marking a significant transition for the robotics industry where he had served as one of its most prominent public advocates.
Reported by The Robot Report
- 14 · Agents & ProductsResolve AI launches SRE-1 model to find production alert causes
Resolve AI has launched SRE-1, an AI model trained to identify the root causes of production alerts, along with Primary On-call, a service that deploys AI agents to handle on-call duties traditionally performed by human engineers.
Reported by SiliconANGLE
- 15 · Funding & DealsNeurotech company leads $400M funding round
This is a weekly roundup of the largest funding rounds announced in the U.S., led by a neurotech company that raised $400 million in Series D funding.
Reported by Crunchbase News
Every story, by topic
One short email a day with the stories that matter. Free. Unsubscribe with one click.