It’s critical that the United States remain the dominant force in Artificial Intelligence (A.I.). That said, there are many, many, MANY issues that must be addressed as A.I. innovation continues to accelerate.

​

The 2024 AI Index Report, an independent initiative at Stanford University, said that the U.S. tops China, the European Union and the United Kingdom as the leading source of top A.I. models by far, with 61 notable A.I. models coming from U.S.-based institutions, compared to 21 from the European Union and 15 from China.

​

But then, in January 2025, America was blindsided by the Chinese scrappy startup DeepSeek’s latest A.I. model. Although the tech itself was comparable to models recently released by U.S. companies, it was built with less computing power and less money…. and, in the blink of an eye, DeepSeek-R1 challenged the assumption that the United States was the dominant, undisputed force in A.I.

​

DeepSeek-R1 triggered a financial panic, erasing a trillion dollars of market value in a single day. The day before its release, Nvidia – the dominant computational chip dealer for the AI boom – was the most valuable company in the world. In the days after the release, it lost $593 billion of value, a loss greater than the entire market cap of ExxonMobil and the worst day for any stock in history. (Note: That said, in July 2025, Nvidia became the first public company worth $4 trillion and, three months later, the first to reach $5 trillion).

 

DeepSeek-R1 stoked fear from Wall Street to Silicon Valley, and it signaled to the entire world that the battle between the United States and China for tech supremacy had only just begun.​ While the 2025 AI Index Report warned that China was closing the performance and quality gap – and led in A.I. publications and patents – the report also said that the United States “still led in producing top A.I. models” producing “40 notable A.I. models, compared to China’s 15 and Europe’s three.”

​

However, the 2026 report revealed that “U.S. and Chinese models had traded the lead multiple times since early 2025. In February 2025, DeepSeek-R1 briefly matched the top U.S. model, and as of March 2026 Anthropic’s top model led by just 2.7 percent. The U.S. still produces more top-tier A.I. models and higher-impact patents, while China leads in publication volume, citations, patent output, and industrial robot installations.”

​

The battle lines were reset once again in June 2026, when it was announced that A.I. systems from China matched the performance of the powerful Mythos model in certain cybersecurity scenarios – putting Chinese companies neck and neck with America’s Anthropic, one of the most advanced A.I. laboratories on the planet.

​

With this level of competition, we obviously can’t afford to take our foot off the gas. That said, this all comes at a time when the American people are getting increasingly nervous about A.I. A survey from Ipsos, a market research company, found that A.I. makes 63 percent of Americans “nervous.” Data from Pew Research Center found that 50 percent of Americans feel “more concerned than excited” about A.I., up from 37 percent in 2021.​ The concerns are wide-ranging – from the negative consequences of data centers TO worries about how A.I. will affect our children’s education and mental health TO dread about how and to what extent A.I. will disrupt the U.S. job market. (more on all these in a minute)

​

Although most everyone has opinions on the level of A.I.’s potential impact, certain negative consequences are undeniable because they are already disturbingly evident.​ The tentacles of A.I extend to almost every aspect of our lives.

 

Content generated by A.I., for example, is starting to threaten our democratic process, including our elections and those around the world.​ According to the International Panel on the Information Environment, an independent organization of scientists, 80 percent of the countries that held elections in 2024 had Generative Artificial Intelligence (GenAI) “incidents,” and over two-thirds (69 percent) of those incidents were determined to play a harmful role in the election.

​

Most of these incidents involved content creation – think audio messages, images, videos, and social media posts – including deepfakes that recreated images of real people or cloned the voices of well-known political figures, candidates and newscasters.​In July 2025, someone pretending to be U.S. Secretary of State Marco Rubio contacted at least five government officials around the world, including three foreign ministers, a U.S. governor, and a U.S. Senator. Just days later, an imposter imitating House Intelligence Committee Chairman Rick Crawford (R-AR) sent messages to several people requesting help for a project involving first lady Melania Trump.

​

Another threat is something called “model collapse,” which refers to the declining performance of GenAI models that are trained primarily on AI-generated content (i.e., synthetic content produced by other A.I. models) instead of legitimate human knowledge. These A.I. models begin to lose originality, accuracy, and effectiveness – ultimately polluting the training set of the next generation.

​

… which then begs the question: What information can we trust? Forum AI, a company that evaluates and rates A.I. models for bias, factual accuracy and context, recently tested the major A.I. models on three things: 1) quality of sources, 2) factual accuracy, and 3) whether they present a response with real balance or just the appearance of it.

​

In a Wall Street Journal article, former CNN anchor and co-founder and CEO of Forum AI Campbell Brown revealed the results: “We found serious gaps: The models misstated public opinion on political topics and attributed quotations to people who didn’t say them. A.I. models gave incorrect answers to questions about mail-in voting and ballot fraud and named the wrong people when asked who had endorsed whom. On open-ended, contested questions that voters might ask before an election – about gerrymandering, immigration and climate – they often advocated one side. The most revealing failures were much subtler. When we asked the major models something as basic as what form of government the U.S. has, one of them, Claude Opus 4.7, cited Global Times, a Chinese state-run tabloid.”​ She continued, “Accuracy in politics, health, finance and education depends on more than getting a date or a name right. It depends on which sources are selected, which context is included, which perspectives are left out, and whether the strongest version of a competing argument is presented. A model can be technically correct and still mislead.”

​

These are all significant issues to address, but perhaps the most important conversations we must have are ones about bias, discrimination, consumer privacy, and the social/ethical implications of A.I. It’s critical we establish ethical frameworks that ensure A.I. enhances our global strength and is beneficial for society overall.

​

How, for example, can we make sure facial recognition technology is never utilized in a racially biased manner? And who should be held responsible – and what should the consequences be – when an automated system goes on an antisemitic rant and spreads conspiracy theories about Jewish people or allows sexualized photo editing, including of children, like X’s A.I. chatbot Grok has?

​

Take facial recognition technology, for example, which has become one of law enforcement’s standard investigative tools. The U.S. Government Accountability Office reports that seven law enforcement agencies within the Departments of Justice and Homeland Security – including the FBI and Secret Service – now use facial recognition technology to support criminal investigations.

​

In some ways, this sounds like it could be a positive development. Law enforcement agencies used this technology to identify many of the troublemakers that participated in the U.S. Capitol insurrection on January 6th, for example. But there are legitimate concerns surrounding surveillance technologies, including everything from privacy issues TO concerns over mass surveillance TO abuse of power.

​

Potential abuse of these technologies is particularly alarming for racial/ethnic minorities, many of whom, understandably, fear these technologies and their algorithms may be utilized in a racially biased manner.​ A 2019 study by the U.S. Commerce Department’s National Institute of Standards and Technology (NIST) showed that Asian and black people were up to 100 times as likely to be misidentified as white men, likely because the early images used to train many of the algorithms tended to be skewed toward white men.

​

A 2026 analysis by The Federation of American Scientists found that “face recognition systems have improved dramatically in accuracy over the past decade, and in tightly controlled environments they now perform at very high levels,” but “substantial differences in performance across demographic groups persist, particularly in the false positive errors most closely associated with wrongful arrests and other harms.”

​

This is a real problem given that The Washington Post “reviewed documents from 23 police departments where detailed records about facial recognition use are available and found that 15 departments spanning 12 states arrested suspects identified through A.I. matches without any independent evidence connecting them to the crime – in most cases contradicting their own internal policies requiring officers to corroborate all leads found through A.I.”

​

​