The Need for AI Regulations, Part III

June 27, 2026

III. Regulating AI

As we have summarized in Section II, the benefits of AI are considerable but so are the risks. The public in Western democracies, the Pope, and the leaders of Anthropic all want to see more government regulation of AI before the risks turn into catastrophes. However, Anthropic stands alone within the Big Tech industry. The other major AI development enterprises have invested a great deal of money into lobbying to combat U.S. regulations. In contrast, the European Union is working to develop a comprehensive framework for AI regulation, with a strong start in the AI Act. A timeline of international AI regulations foreseen in October 2022, but not enacted in all cases, is shown in Fig. III.1. After reviewing the current situation in the U.S. and EU, we will suggest some possible regulations that could serve to seriously mitigate the risks, along with what additional features users should demand of AI models.

Most of the AI industry has viewed government regulation as an unnecessary impediment (or more dramatically “a death blow”) to ongoing innovation and the U.S. competitive role vis-à-vis China and other countries. Starting in earnest in 2024 various Big Tech companies and venture capital groups have invested large sums of money in political action committees (Super PACs) created explicitly to lobby the U.S. federal and state governments against AI regulations. Meta (the parent company of Facebook and Instagram) invested tens of millions in 2025 in the Super PAC American Technology Excellence Project to support tech-friendly political candidates and to oppose emerging state regulations. The venture capitalist firm Andreessen Horowitz and OpenAI president Greg Brockman invested $100 million in the super PAC Leading the Future to lobby strongly against strict AI regulation. Meta, Microsoft, Amazon, and Alphabet (Google) all lobbied strongly in favor of inclusion within the 2025 Budget Reconciliation Act (or “One Big Beautiful Bill Act”) of an explicit 10-year moratorium on state laws regulating AI. That moratorium was passed by the U.S. House of Representatives but subsequently removed from the final bill by the Senate.

The only AI-related bill that has to date passed both houses of the U.S. Congress and been signed into law is the 2025 Take It Down Act, which bans non-consensual intimate images, including those created by AI deepfakes. The act creates federal penalties for creating such images and requires social media platforms to take them down once alerted to their existence. Under discussion in the current Congress in 2026 is the Great American AI Act which, despite its self-aggrandizing title, is mostly industry-friendly. It would aim to preempt many, but not all, state AI regulation laws and to increase penalties for fraud carried out with the aid of AI. The bill “would formalize the Center for AI Standards and Innovation (CAISI) within the Department of Commerce and assign it a central role in AI evaluation, standards development, incident reporting, independent verification, and federal coordination. CAISI would develop voluntary guidelines and best practices for AI security, interpretability, evaluation, synthetic content detection, cyber incident response, and certain national security safeguards, among others.“ Furthermore, this bill “is primarily aimed at the biggest AI developers with more than $500 million in revenue that are building cutting-edge AI models rather than most typical businesses developing in-house AI or deploying commercial AI models.”

In contrast the EU AI Act goes well beyond voluntary guidelines. It establishes a risk-based approach, illustrated in Fig. III.2, that applies different levels of regulation to AI systems that pose different levels of risk. Systems in the highest-risk category — for example, involving social scoring systems (assigning scores to individuals based on their personal traits and appearance) or efforts to manipulate users by exploiting deception or vulnerabilities — are strictly prohibited. Also prohibited are attempts to create facial recognition databases “by untargeted scraping of facial images from the internet or CCTV footage” and real-time biometric identification in public spaces, except when necessary and documented as essential for solving or preventing crimes.

AI systems judged to be in the second tier of “high-risk” must undergo a conformity assessment. The systems covered include ones covered by product safety regulations (e.g., medical devices, vehicles, toys), ones used for biometric identification and categorization, ones that affect access to essential services, ones that are used in educational and vocational training, and ones used in management of workers and employment. AI systems in this high-risk category are required in the law to:

• establish risk and quality management systems,
• to allow users to implement human oversight,
• to demonstrate that “training, validation and testing datasets are relevant, sufficiently representative and, to the best extent possible, free of errors and complete according to the intended purpose,”
• to achieve appropriate levels of accuracy, robustness, and cybersecurity,
• to provide technical documentation and instructions for use so that deployers can also comply with the law,
• and to keep records of system modifications and of events relevant for identifying national level risks.

General-purpose AI (GPAI) systems, including the LLM chatbots and GenAI models we have discussed in Section II, fall within tiers 3 and 4. All of them are required by the EU AI Act to:

• Draw up technical documentation, including training and testing process and evaluation results.
• Draw up information and documentation to supply to downstream providers that intend to integrate the GPAI model into their own AI system in order that the latter understands capabilities and limitations and is enabled to comply.
• Establish a policy to respect the Copyright Directive.
• Publish a sufficiently detailed summary about the content used for training the GPAI model.

GPAI systems that exceed a specified threshold of computing intensity are deemed to pose systemic risk and have additional obligations to perform and document model evaluations, including adversarial testing to assess and mitigate systemic risk, and to track, document, and report serious incidents that arise in use of their systems.

AI systems deemed to pose minimal risk fall in tier 4, for which there are no mandatory obligations. Rather the EU AI Act encourages voluntary codes of conduct to promote responsible AI development.

What U.S. and other governments should regulate:

The EU AI Act encompasses comprehensive regulations. For the U.S. Congress to come to bipartisan agreement on comparably comprehensive regulations would likely take so many years that the legislation would be moot by the time it passed. Below we offer our own suggestions on individual government regulations that may make sense to seek on specific aspects of AI, a number of which are incorporated in the EU AI Act. Some of the general considerations that should guide regulations are illustrated in Fig. III.3.

• Truth in labeling: The capability with the latest GenAI models to produce realistic, convincing deepfake images and videos creates unacceptable risks to cause panic, market crashes, wars, defamation, personal injury, and societal breakdowns. These risks must be addressed by requiring the GenAI models to include unerasable labelling (e.g., “AI-generated” or “Generated by AI model …”) via watermarks or equivalent on every image and video they generate. Users who create art with GenAI models could use these labels as part of their signature. Avoidance or removal of such labels should be subject to prosecution and penalties for fraud.
• Prohibition of voice cloning: In analogy with the Take It Down Act, the use of AI to clone other people’s voices without prior consent of the voice owner should be banned and subject to criminal penalties.
• Transparency about training: AI providers should be required to make available on publicly accessible websites details about what materials, including what copyrighted materials, are included in the datasets they use to train the models. They should also provide additional documentation reporting on the testing, validation, and risk assessment procedures followed and results obtained prior to public release of the model. This information provides users with some ability to judge the models for possible bias and inaccuracy. 
• Consumer Choice: At present, users of the Google search engine are automatically provided with a list of AI-generated content.  Users should be allowed to choose whether or not they want to be supplied with content generated through AI, given the considerably amount of energy required to provide an AI answer compared to traditional Google searches. 
• Independent oversight board: An independent oversight board should be established to review performance and possible vulnerabilities in new generations of AI code, using the documentation made available by the providers under the transparency requirement.  There are numerous examples of government agencies that regulate similar important segments of society.  These examples would include the Federal Aviation Agency (FAA) that regulates air transportation; the Food and Drug Administration (FDA) that regulates food and drug industries; and the Federal Communications Commission (FCC) that regulates communications media. 
• Copyright requirements: AI providers should be required to license any non-public copyrighted material they use in training their models. In answering user queries the models should alert users to the use of copyrighted materials in the answers provided. Compliance with such a requirement will be quite cumbersome, but without this requirement all copyright law becomes moot because any AI user could copy material from AI answers without awareness of the intellectual property rights of the originator.
• Resource allowances: Federal regulatory agencies should establish allowance limits for each major AI provider on the amount of fossil-fuel-generated electricity, in all forms, and water that the company can use annually for provider’s staff, equipment, and all data centers, with fines assessed for exceeding those limits.  As part of this suggestion, we recommend that states place a moratorium on tax breaks awarded for new data centers.  Such centers use immense amounts of electricity and water; they employ very small staff once they are constructed; many of them depend on fossil fuel generation of energy; and they are a source of significant noise pollution. 
• Liability: Establish liability for AI companies when their software is used for fraud, defamation, invasion of privacy, or criminal activities, when those applications could have been readily prevented by restrictions built into the software, especially if such restrictions had been previously recommended by the independent oversight board.  This will require repealing Section 230 of the Communications Decency Act, which currently absolves internet providers of liability for nearly anything that they publish. 

Over and above government regulations, we feel there are some qualities that users should demand of AI models they use:

• “Humility”: there are some queries – regarding physical, emotional, or mental harm to a user or others, producing defamatory product, or criminal violations of existing laws – that must not be answered. This requires AI models to become clever enough to recognize when a user is requesting such information in an indirect way, for example, for use in a story or movie the user claims to be creating. In addition, AI generators must admit when they cannot find a reliable answer in the database or when they are “guessing” at an answer or “creating” information.
• “Bullshit detection”: AI programs must make a serious effort to distinguish information with a firm basis in reality from misinformation and disinformation, by checking sources for well-known argument flaws and consistency within the source and with well-established external information. When there is reason for doubt, AI models must acknowledge that doubt in answers to user queries. As we have noted in our previous post about Elon Musk’s AI platform Grok, providers who do not make such efforts can contribute, whether knowingly or unwittingly, to the viral spread of misinformation.
• Transparency: Detailed information about the sources used in training datasets in AI programs must be available and up-to-date on provider websites, so that users can consider any possible bias that may exist in the training.
• Limits on use of user data: Users should be able to refuse to provide information on their income, political leanings, personal situation, or other private data, except insofar as that information may be revealed in questions they pose to AI.
• Public reporting on resource usage: AI companies must make available to the public annually the total amount of energy (including energy sources) and water usage by their data centers.

We reject the idea promulgated by some in Silicon Valley that any attempt to place constraints on AI will cause a “death blow to innovation.” Innovation and safety must be given equal weight in advancing the technology. And history provides pretty strong evidence that industries left to their own devices for protecting user safety generally place safety concerns well below providing short-term profit to their owners and shareholders. If AI advancement slows down somewhat by being regulated to advance responsibly, it will produce benefits that strongly outweigh its risks.

References:

G. Marcus, Taming Silicon Valley: How We Can Ensure that AI Works for Us (The MIT Press, 2024), https://www.amazon.com/Taming-Silicon-Valley-Ensure-Works-ebook/dp/B0CQWWM94N/ref=sr_1_1

A. Smith, Study Shows State and Local Opposition to New Data Centers is Gaining Steam, NBC News, Nov. 14, 2025, https://www.nbcnews.com/politics/economics/state-local-opposition-new-data-centers-gaining-steam-rcna243838

L. DePillis, Local Opposition is Slowing A.I. Data Centers. Wall Streed has Noticed, New York Times, Mar. 26, 2026, https://www.nytimes.com/2026/03/26/business/economy/ai-data-centers-construction-local-opposition.html

M. Dorgan, Agitators United by Chinese Money, Hate for America Target Data Centers, Experts Warn, Fox News, May 24, 2026, https://www.foxnews.com/politics/agitators-united-chinese-money-hate-america-target-data-centers-experts-warn

DebunkingDenial, The Cyberhogs: Cryptocurrency Mining, Artificial Intelligence, and Data Centers, https://debunkingdenial.com/portfolio/the-cyberhogs-cryptocurrency-mining-artificial-intelligence-and-data-centers/  

Epoch AI, How Much Does it Cost to Train Frontier AI Models?, June 3, 2024, https://epoch.ai/publications/how-much-does-it-cost-to-train-frontier-ai-models

S. Guthrie, Inside the World’s Most Powerful AI Data Center, Microsoft Blog, Sep. 18, 2025, https://blogs.microsoft.com/blog/2025/09/18/inside-the-worlds-most-powerful-ai-datacenter/

S. Sumu, What are 10 Largest AI Data Centers in the World?, Tech Chilli, Dec. 15, 2025, https://techchilli.com/artificial-intelligence/largest-ai-data-center-in-the-world/

M. Honan, Data Centers are Amazing. Everyone Hates Them, MIT Technology Review, Jan. 14, 2026, https://www.technologyreview.com/2026/01/14/1131253/data-centers-are-amazing-everyone-hates-them/

S. Chen, How Much Energy Will AI Really Consume? The Good, the Bad and the Unknown, Nature, Mar. 5, 2025, https://www.nature.com/articles/d41586-025-00616-z

Imagera Research Team, AI Image Generation Statistics 2026: 80 Million Per Day, June 12, 2026, https://imagera.ai/blog/ai-image-generation-statistics-2026

A. Devasia, 25+ AI Data Center Statistics & Trends, The Network Installers, Oct. 8, 2025, https://thenetworkinstallers.com/blog/ai-data-center-statistics/

T. Mills, Re-Architecting the AI Server: The Hidden Water Cost of Data Centers (Part II), Harvard Science Review, Feb. 28, 2026, https://harvardsciencereview.org/2026/02/28/re-architecting-the-ai-server-the-hidden-water-cost-of-data-centers-part-ii/

C.D. Walker and I. Goldsmith, From Energy Use to Air Quality, the Many Ways Data Centers Affect US Communities, World Resources Institute, Feb. 17, 2026, https://www.wri.org/insights/us-data-center-growth-impacts

A. Hassan, The Cloud Has Sound: The Unrelenting and Unseen Cost of A.I. Data Centers, New York Times, June 22, 2026, https://www.nytimes.com/2026/06/17/us/data-centers-noise-pollution.html

C. Krittanawong, et al., PM_2.5 and Cardiovascular Diseases: State-of-the-Art Review, International Journal of Cardiology, Cardiovascular Risk and Prevention 19, 200217 (2023), https://www.sciencedirect.com/science/article/pii/S2772487523000508

L. Mineo, Why Are Communities Pushing Back Against Data Centers?, The Harvard Gazette, Apr. 9, 2026, https://news.harvard.edu/gazette/story/2026/04/why-are-communities-pushing-back-against-data-centers/

Voronoi, Electricity Use from AI Data Centers Will Quadruple by 2030, May 10, 2025, https://www.voronoiapp.com/energy/Electricity-Use-From-AI-Data-Centers-Will-Quadruple-by-2030-5038

M. Joselow and B. Plumer, Trump Offers Funds for First New U.S. Coal Plants in 13 Years, New York Times, June 4, 2026, https://www.nytimes.com/2026/06/04/climate/trump-coal-plants-funding.html

Y.Y. Chung, F. Darakdjian, and T. Leahy, When AI Meets Water Scarcity: Data Centers in a Thirsty World, MSCI, Dec. 9, 2025, https://www.msci.com/research-and-insights/blog-post/when-ai-meets-water-scarcity-data-centers-in-a-thirsty-world

Global Water Consumption by Generative Artificial Intelligence, AIdrinkswater.com, https://www.aidrinkswater.com/report.html

J. Mecke, DeepSeek’s AI Training Costs Exposed: What Early-Stage SaaS CEOs Must Know in 2025, Development Corporate Management Consulting, Sep. 21, 2025, https://developmentcorporate.com/product-management/deepseeks-ai-training-costs-exposed-what-early-stage-saas-ceos-must-know-in-2025/

P. Hanbury, J. Wang, P. Brick, and A. Cannarsi, DeepSeek: A Game Changer in AI Efficiency?, Bain & Company, Feb. 4, 2025, https://www.bain.com/insights/deepseek-a-game-changer-in-ai-efficiency/

V. Sukumar, Shifting AI Inference from the Cloud to Your Phone Can Reduce AI Costs, Qualcomm, Sep. 16, 2025, https://www.qualcomm.com/news/onq/2025/09/shifting-ai-inference-from-the-cloud-to-your-phone-can-reduce-ai-costs

 N. Gillespie, et al., Trust, Attitudes and Use of Artificial Intelligence: A Global Study 2025, The University of Melbourne and KPMG, https://mbs.edu/-/media/PDF/Research/Trust_in_AI_Report.pdf

Annenberg Public Policy Center, Many Americans Pessimistic About AI’s Impact – and Want More Regulation, May 13, 2026, https://www.annenbergpublicpolicycenter.org/americans-pessimistic-about-ais-impact-and-want-more-regulation/

I. Piro, Pope Leo’s ‘Magnifica Humanitas’: AI Must Serve Humanity Not Concentrate Power, Vatican News, May 2026, https://www.vaticannews.va/en/pope/news/2026-05/pope-leo-xiv-encyclical-magnifica-humanitas-ai.html

C. Murray, Anthropic Official Will Join Pope Leo for New AI Church Policy Statement, Forbes, May 18, 2026, https://www.msn.com/en-us/technology/artificial-intelligence/anthropic-official-will-join-pope-leo-for-new-ai-church-policy-statement/ar-AA23ugBp

Pew Research Center, How the U.S. Public and AI Experts View Artificial Intelligence, Apr. 3, 2025, https://www.pewresearch.org/internet/2025/04/03/how-the-us-public-and-ai-experts-view-artificial-intelligence/

Taylor Swift Deepfakes Spread Online, Sparking Outrage, CBS News, Jan. 26, 2024, https://www.cbsnews.com/news/taylor-swift-deepfakes-online-outrage-artificial-intelligence/

E. Saslow, The World’s Leading Deepfake Expert No Longer Trusts His Own Eyes, New York Times Magazine, June 14, 2026, https://www.nytimes.com/2026/06/14/us/ai-deepfake-hany-farid.html

Wikipedia, TAKE IT DOWN Act, https://en.wikipedia.org/wiki/TAKE_IT_DOWN_Act  

K. Conger, D. Freedman, and S.A. Thompson, Musk’s Chatbot Flooded X with Millions of Sexualized Images in Days, New Estimates Show, New York Times, Jan. 22, 2026, https://www.nytimes.com/2026/01/22/technology/grok-x-ai-elon-musk-deepfakes.html

M. Lazzouni, Voice Cloning: What It Is and Why It’s Scary, BuiltIn, June 14, 2023, https://builtin.com/artificial-intelligence/what-is-voice-cloning

G. Contino, The Voice Scam Call Portrayed in ‘Thelma’ is Real and an Increasing Threat in the Age of AI, CNBC, July 5, 2024, https://www.cnbc.com/2024/07/05/ai-voice-scam-call-movie-thelma-growing-threat.html

R. Bellan, Leaked Meta AI Rules Show Chatbots Were Allowed to Have Romantic Chats with Kids, TechCrunch, Aug. 14, 2025, https://techcrunch.com/2025/08/14/leaked-meta-ai-rules-show-chatbots-were-allowed-to-have-romantic-chats-with-kids/

https://www.sugarlab.ai/ai-pornstar-generator

Lawsuit Accuses ChatGPT of Helping Gunman Plan FSU Mass Shooting, PBS News, May 11, 2026, https://www.pbs.org/newshour/nation/lawsuit-accuses-chatgpt-of-helping-gunman-plan-fsu-mass-shooting

T. Farrant, An AI Agent Deleted a Company’s Entire Database in 9 Seconds – Then Wrote an Apology, Euro News, Apr. 28, 2026, https://www.euronews.com/next/2026/04/28/an-ai-agent-deleted-a-companys-entire-database-in-9-seconds-then-wrote-an-apology

K. Roose, Anthropic Claims Its New A.I. Model, Mythos, is a Cybersecurity ‘Reckoning’, New York Times, Apr. 7, 2026, Anthropic Claims Its New A.I. Model, Mythos, Is a Cybersecurity ‘Reckoning’ – The New York Times

Wikipedia, Zero-Day Vulnerability, https://en.wikipedia.org/wiki/Zero-day_vulnerability  

World Economic Forum, Global Cybersecurity Outlook 2025, https://reports.weforum.org/docs/WEF_Global_Cybersecurity_Outlook_2025.pdf

D. Volz, Google Says Criminal Hackers Used A.I. to Find a Major Software Flaw, New York Times, May 11, 2026, https://www.nytimes.com/2026/05/11/us/politics/google-hackers-attack-ai.html

B. Gil, Meta Says Thousands of Instagram Accounts Were Breached Through Its AI Support Assistant, Gizmodo, June 8, 2026, https://gizmodo.com/meta-says-thousands-of-instagram-accounts-were-breached-through-its-ai-support-assistant-2000768770

R. Min, AI Models Can Hack Computers and Self-Replicate Onto New Machines, New Research Finds, Euro News, May 9, 2026, https://www.euronews.com/next/2026/05/09/ai-models-can-hack-computers-and-self-replicate-onto-new-machines-new-research-finds

A. Capoot, Anthropic and U.S. Government to Face Off in DC Court over Blacklisting of AI Company, CNBC, May 19, 2026, https://www.cnbc.com/2026/05/19/anthropic-dod-blacklist-court-opening-arguments.html

R. Knappenberger, DC Circuit Slams Pentagon Blacklisting of Anthropic as Overreach, Courthouse News Service, May 19, 2026, https://www.courthousenews.com/dc-circuit-slams-pentagon-blacklisting-of-anthropic-as-overreach/

M. O’Brien, Trump Signs an Executive Order that Invites Vetting of Top AI Models for National Security Risks, AP News, June 2, 2026, https://apnews.com/article/trump-ai-executive-order-e41af74f7b0865482f07d10fe7a50fe3

Wikipedia, Machine Learning, https://en.wikipedia.org/wiki/Machine_learning

Stanford University Human-Centered Artificial Intelligence, The 2026 AI Index Report, https://hai.stanford.edu/ai-index/2026-ai-index-report

A. Park, Anthropic Billionaire Co-Founder Joins Pope Leo, Warns AI Job Losses Will Spark “Moral Imperative of Historic Proportions”, Forbes, May 25, 2026, https://www.forbes.com/sites/aliciapark/2026/05/25/anthropic-billionaire-cofounder-joins-pope-leo-warns-ai-job-losses-will-spark-moral-imperative-of-historic-proportions/

Wikipedia, Deep Blue versus Garry Kasparov, https://en.wikipedia.org/wiki/Deep_Blue_versus_Garry_Kasparov

BBC News, Deep Blue vs. Kasparov: How a Computer Beat Best Chess Player in the World, https://www.youtube.com/watch?v=KF6sLCeBj0s

Wikipedia, AlphaGo versus Lee Sedol, https://en.wikipedia.org/wiki/AlphaGo_versus_Lee_Sedol  

O. Vinyals, et al., Grandmaster Level in StarCraft II Using Multi-Agent Reinforcement Learning, Nature 575, 350 (2019), https://www.nature.com/articles/s41586-019-1724-z

M. Jaderberg, et al., Human-Level Performance in 3D Multiplayer Games with Population-Based Reinforcement Learning, Science 364, 859 (2019), https://www.science.org/doi/10.1126/science.aau6249

DARPA, AlphaDogfight Trials Go Virtual for Final Event, Aug. 7, 2020, https://www.darpa.mil/news/2020/alphadogfight-trials-final

Wikipedia, Generative Adversarial Network, https://en.wikipedia.org/wiki/Generative_adversarial_network

Wikipedia, DALL-E, https://en.wikipedia.org/wiki/DALL-E

Northwestern University Buffett Brief, The Rise of Artificial Intelligence and Deepfakes, Sep. 2025, https://buffett.northwestern.edu/documents/buffett-brief_the-rise-of-ai-and-deepfake-technology.pdf

Wikipedia, Taylor Swift Deepfake Pornography Controversy, https://en.wikipedia.org/wiki/Taylor_Swift_deepfake_pornography_controversy

S.A. Thompson, T.M. Terol, K. Conger, and D. Freedman, How Elon Musk is Remaking Grok in His Image, New York Times, Sep. 2, 2025, https://www.nytimes.com/2025/09/02/technology/elon-musk-grok-conservative-chatbot.html

DebunkingDenial, Can You Create an ‘Evil’ AI? Let’s Ask Grok, https://debunkingdenial.com/can-you-create-an-evil-ai-lets-ask-grok/

GitHub Copilot, https://github.com/features/copilot

K. Mao, Sapienz: Intelligent Automated Software Testing at Scale, Meta Engineering, May 2, 2018, https://engineering.fb.com/2018/05/02/developer-tools/sapienz-intelligent-automated-software-testing-at-scale/

Anthropic, Project Glasswing, https://www.anthropic.com/glasswing

A. Capoot, Anthropic Disables Access to Fable 5 and Mythos 5 to Comply with Government Directive, CNBC, June 12, 2026, https://www.cnbc.com/2026/06/12/anthropic-disables-access-to-fable-5-and-mythos-5-to-comply-with-government-directive.html

Crowdstrike, Crowdstrike 2026 Global Threat Report, https://go.crowdstrike.com/2026-global-threat-report.html   

Lindau Nobel Laureate Meetings, Nobel Prize in Chemistry 2024: Protein Folding and AI, https://www.lindau-nobel.org/blog-nobel-prize-in-chemistry-2024-protein-folding-and-ai/

T. Hirosawa and T. Shimizu, A Narrative Review of Artificial Intelligence in Medical Diagnostics, Computers, Materials and Continua 83, 3919 (2025), https://www.sciencedirect.com/org/science/article/pii/S1546221825004874

NVIDIA, https://www.nvidia.com/en-us/

3Blue1Brown, Large Language Models Explained Briefly, https://www.youtube.com/watch?v=LPZh9BOjkQs

Wikipedia, GPT-5, https://en.wikipedia.org/wiki/GPT-5

Reddit, Sam Altman Won’t Tell You That GPT-4 Has 220B Parameters, https://www.reddit.com/r/LocalLLaMA/comments/14p26g6/sam_altman_wont_tell_you_that_gpt4_has_220b/

R.L. Draelos, et al., Large Language Models Provide Unsafe Answers to Patient-Posed Medical Questions, Nature NPJ Digital Medicine 9, Article #241 (2026), https://www.nature.com/articles/s41746-026-02428-5

OpenAI, GPT-4 System Card, https://cdn.openai.com/papers/gpt-4-system-card.pdf

J. Foulds, P. Feldman, and S. Pan, Grok’s Antisemitic Rant Shows How Generative AI Can Be Weaponized, The Conversation, July 9, 2025, https://theconversation.com/groks-antisemitic-rant-shows-how-generative-ai-can-be-weaponized-257880

L. Hagen, H. Jingnan, and A. Nguyen, Elon Musk’s AI Chatbot, Grok, Started Calling Itself ‘MechaHitler’, NPR, July 9, 2025, https://www.npr.org/2025/07/09/nx-s1-5462609/grok-elon-musk-antisemitic-racist-content

G. Lewis-Kraus, What Is Claude? Anthropic Doesn’t Know, Either, The New Yorker, Feb. 9, 2026, https://www.newyorker.com/magazine/2026/02/16/what-is-claude-anthropic-doesnt-know-either

GeniusAITech, AI Statistics 2025: Market Growth, Usage, Trends & Adoption, https://geniusaitech.com/ai-statistics-2025/

McKinsey Global Institute, Generative AI and the Future of Work in America, July 26, 2023, https://www.mckinsey.com/mgi/our-research/generative-ai-and-the-future-of-work-in-america

World Economic Forum, The Future of Jobs Report 2023, https://www.weforum.org/publications/the-future-of-jobs-report-2023/digest/

World Economic Forum, Future of Jobs Report 2025, https://reports.weforum.org/docs/WEF_Future_of_Jobs_Report_2025.pdf

Wikipedia, AI Slop, https://en.wikipedia.org/wiki/AI_slop

O. Whang, Can You Get Rich Quick Off A.I. Slop?, New York Times Magazine, June 1, 2026, https://www.nytimes.com/interactive/2026/06/01/magazine/ai-slop-viral-videos.html

S. Cole, ‘Life or Death’: AI-Generated Mushroom Foraging Books are All Over Amazon, 404 Media, Aug. 29, 2023, https://www.404media.co/ai-generated-mushroom-foraging-books-amazon/

White House, The MAHA Report, May 2025,  https://www.whitehouse.gov/wp-content/uploads/2025/05/MAHA-Report-The-White-House.pdf

L. Tuquero, How Fake Citations Appeared in RFK Jr.’s MAHA Report: Here are Generative AI’s Red Flags in Studies, PolitiFact, May 30, 2025, https://www.politifact.com/article/2025/may/30/MAHA-report-AI-fake-citations/

E. Temple, Nobel Laureate Olga Tokarczuk Apparently Used AI to Write Her Latest Novel, Literary Hub, May 19, 2026, https://lithub.com/nobel-laureate-olga-tokarczuk-apparently-used-ai-to-write-her-latest-novel/

Arena, https://arena.ai/

Floyd Bishop, Deepfake Challenge 2025, https://www.youtube.com/watch?v=FERa1AI2EK8

GetReal Security,  https://www.getrealsecurity.com/

DebunkingDenial, How to Tune Your Bullshit Detector, https://debunkingdenial.com/how-to-tune-your-bullshit-detector-part-i/  

B.D. Earp, The Unbearable Asymmetry of Bullshit, Quillette, Feb. 15, 2016, https://quillette.com/2016/02/15/the-unbearable-asymmetry-of-bullshit/

T. Farley, X Post, May 29, 2014, https://x.com/krelnik/status/472046082135162881

M.M. Grynbaum and R. Mac, The Times Sues OpenAI and Microsoft Over A.I. Use of Copyrighted Work, New York Times, Dec. 27, 2023, https://www.nytimes.com/2023/12/27/business/media/new-york-times-open-ai-microsoft-lawsuit.html

G. Marcus and R. Southen, Generative AI Has a Visual Plagiarism Problem, IEEE Spectrum, Jan. 6, 2024, https://spectrum.ieee.org/midjourney-copyright

Reid Southen, https://www.reidsouthenart.com/

DebunkingDenial, Artificial Intelligence: Is It Smart? Is It Dangerous?, https://debunkingdenial.com/artificial-intelligence-is-it-smart-is-it-dangerous-part-ii-chatbots-and-the-path-to-artificial-general-intelligence/

DebunkingDenial, Test Your Critical Thinking, https://debunkingdenial.com/test-your-critical-thinking-part-i-questions/

MMMU, A Massive Multi-Discipline and Multimodal Understanding and Reasoning Benchmark for Expert AGI, https://mmmu-benchmark.github.io/

D. Rein, et al., GPQA: A Graduate-Level Google-Proof Q&A Benchmark, https://arxiv.org/pdf/2311.12022  

Center for AI Safety, Scale AI, and HLE Contributors Consortium, A Benchmark of Expert-Level Academic Questions to Assess AI Capabilities, Nature 649, 1139 (2026), https://www.nature.com/articles/s41586-025-09962-4

A. Safar, ClockBench: Visual Time Benchmark Where Humans Beat the Clock, LLMs Don’t, https://clockbench.ai/ClockBench.pdf

DebunkingDenial, Changing Minds in a Polarized Era, https://debunkingdenial.com/changing-minds-in-a-polarized-era/

T. Eldho Alias, From Language Models to General Intelligence: How Close is the World to AGI?, Forbes, July 2, 2025, https://www.forbes.com/councils/forbestechcouncil/2025/07/02/from-language-models-to-general-intelligence-how-close-is-the-world-to-agi/

I. Belcic and C. Stryker, What Is Catastrophic Forgetting?, IBM, https://www.ibm.com/think/topics/catastrophic-forgetting

H. Taneja, The Era of “Move Fast and Break Things” is Over, Harvard Business Review, Jan. 22, 2019, https://hbr.org/2019/01/the-era-of-move-fast-and-break-things-is-over

T. Theoto, et al., Responsible AI for an Era of Tighter Regulations, BCG, Oct. 10, 2022, https://www.bcg.com/publications/2022/acting-responsibly-in-tight-ai-regulation-era

C. Lusinchi, AI Regulations in US: Understanding the Roadblocks and Future Pathways, Nemko Digital, Dec. 1, 2025, https://digital.nemko.com/insights/how-big-tech-lobbying-stopped-us-ai-regulation-in-2025

M. Brown and M. O’Brien, House Republicans Include a 10-Year Ban on US States Regulating AI in ‘Big Beautiful’ Bill, AP News, May 16, 2025, https://apnews.com/article/ai-regulation-state-moratorium-congress-39d1c8a0758ffe0242283bb82f66d51a

JDSupra, The Great American AI Act: What It Means – and Doesn’t Mean – for Companies Using AI, June 11, 2026,  https://www.jdsupra.com/legalnews/the-great-american-ai-act-what-it-means-7369971/

Nemko Digital, The EU AI Act, https://digital.nemko.com/regulations/eu-ai-act

EU Artificial Intelligence Act, High-Level Summary of the AI Act, Feb. 27, 2024, https://artificialintelligenceact.eu/high-level-summary/

Women in AI & Robotics, AI Regulation, Feb. 28, 2022,  https://www.womeninairobotics.de/post/ai-regulation