Open weight models are communist, allegedly. More scary stories. New York’s data centre moratorium. Hollywood absurdities. Data centre protests, and a kill-switch for AI.

Welcome to a new edition of misaligned bits, the (roughly weekly) newsletter from Misaligned where we sum up recent news and research, sometimes with a lighter touch.

As usual, we will mark all non-medium links with “➚” (external link) and all possibly paywalled links with “🔒”.

AI Communism?

A few weeks ago we reported that Dean Ball, a former top White House official of the Trump administration and main author of Trump’s “AI Action Plan”, has joined OpenAI as “as head of strategic futures”. While we are still not sure what that job description exactly means, he has certainly gone to work: In a lengthy X/Twitter post he explains that open weight models that are freely available for download are a sign of “AI communism”. Ball apparently sees “AI for public good” as a “dystopian hellscape”. (➚Ball on X)

“One probable outcome of an open-weight-model-dominant world is full AI communism” he writes. “[…] rather than a market product, AI is a ‘public good’ which will ultimately be provided by the state as a kind of ‘digital public infrastructure.’ This future strikes me as a dystopian hellscape […]”.

He furthermore hints at the Trump administration banning such open weight models: “I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models.”

That would, of course, guarantee the US AI industry a desperately needed justification for their high-priced services. May that all as it be, it appears that OpenAI has hired Ball to move them closer to the Trump administration’s idea of protectionism, which is certainly one possible future, strategically. It also appears that OpenAI is bringing back the “red scare” accordingly.

Other “scary” stuff

Meanwhile, OpenAI has claimed that an “autonomous AI agent” powered by its AI technology has gone rogue during a test and hacked prominent startup Hugging Face “by itself” in what OpenAI calls an “unprecedented incident” (➚The Guardian) As actual details of the incident are rare one wonders what is worse: That what OpenAI claims is true, or that OpenAI is simply seeking attention as the waning public interest in ChatGPT causes them to place another scare story in the media (which largely appears to uncritically report it as fact).

A scare of the more economic kind was delivered by Dario Amodei, the CEO of Anthropic, who warned that if AI growth forecasts are off by just a year, ‘then you go bankrupt’ (➚Fortune). Well, there’s better not be a bubble then.

US bits

Alarmed by all the scary stories, a new bipartisan proposal in Washington is meant to address risks posed by advanced AI models by mandating a “kill switch” (➚AI Kill Switch Act). The bipartisan “AI Kill Switch Act” would give the Department of Homeland Security the authority to order top artificial intelligence firms to shut down or slow AI models that the government deems too dangerous (➚Politico).

Data centres

New York has become the first U.S. state to halt construction ​of large new data centres, imposing a one-year moratorium. Governor Kathy Hochul said that “as data center development threatens to hike up utility bills, deplete our natural ​resources, and create uncertainty for New Yorkers, it’s my responsibility to take action and lead”. The ban will apply to data centres that use 50 megawatts ⁠or more of power. (➚Reuters)

Meanwhile, Opponents of the data centres build out held ​142 protests across 42 states in the US last Saturday in the first nationwide protest against the AI infrastructure expansion (➚Reuters).

British bits

The new UK prime minister Andy Burnham has replaced the Department for Science, Innovation and Technology with a department and ministry dedicated to AI (➚BBC) and named Kanishka Narayan as “AI minister”.

His responsibilities reportedly will span “AI opportunities, the AI Security Institute, semiconductors and online safety”. However, so far it does not appear that this new department will have a dedicated regulatory role, something that the UK is still missing.

Hollywood absurdities

While George Lucas is claiming that AI in movies will be inevitable, and we should get used to it, Elon Musk has announced he will produce a “historically accurate” version of Homer’s Odyssey by end of the year.

Musk, who has used his social media leverage to campaign against the new Christopher Nolan blockbuster because of its diverse cast, wants to use his AI video generator Grok to produce the movie (➚Hollywood Reporter).

While it is not clear how a “historically accurate” cyclops would look like, we have to assume what Musk really means by “historically accurate” is “white”. If Musk plans to bring his non divers AI slop to the IMAX screen is yet unknown.

Social Media Addiction

A much anticipated trial about social media addiction has been called off at the last minute after the plaintiff settled with the case with YouTube, TikTok and Snap, and has now dropped the case against meta (➚CNN).

Science Bits

Novel bias

A study by Wu et al. demonstrates that “LLMs can spontaneously develop novel social biases about artificial demographic groups even when no inherent differences exist”.

The study finds that “these biases result in highly stratified task allocations, which are less fair than assignments by human participants and are exacerbated by newer and larger models”.

The authors state that “Bias in large language models (LLMs) is becoming an even more urgent problem as these models are put into agents that make decisions” and “that the current way that we focus on removing biases from models is not enough.”

The study concludes that “these results reveal that LLMs are not merely passive mirrors of human social biases, but can actively create new ones from experience, raising urgent questions about how these systems will shape societies over time.”

Wu, A. J., Liu, R., Bai, X., & Griffiths, T. L. (2026). Large Language Models Develop Novel Social Biases Through Adaptive Exploration. Forty-Third International Conference on Machine Learning

AI in education and reduced skill acquisition

While AI is being pushed more and more into education, we look at a study published last year. The paper “examines the impact of generative AI […] on student learning in math education.”

The study finds “that although AI-based tutoring improves performance during practice sessions, students relying on the technology may underperform when access to AI is subsequently removed, indicating reduced skill acquisition.”

It concludes that “when access is subsequently taken away, students actually perform worse than those who never had access (17% reduction in grades for GPT Base) — i.e., unfettered access to GPT-4 can harm educational outcomes.”

H. Bastani, O. Bastani, A. Sungu, H. Ge, Ö. Kabakcı, & R. Mariman (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics, Proc. Natl. Acad. Sci. U.S.A. 122 (26) e2422633122

Follow and subscribe

If you stumbled across this article on the web, subscribe to the “misaligned bits” newsletter and follow Misaligned on Medium, LinkedIn, Threads or Mastodon.

You can now also subscribe to this newsletter directly with your email.