Nobody likes data centres, MIT reports, German court rules on Google. And how oppressive are AI systems?

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 “🔒”.

Polling bits

A majority of teachers show concern about the impact of artificial intelligence (AI) on students’ critical thinking skills, according to a new NPR/Ipsos poll. More than half of the teachers polled view AI as mostly a shortcut for students to avoid doing more work (55%) and say it is making it harder for students to learn critical thinking skills (54%) (➚Ipsos).

Meanwhile, the Brookings Institute has published a comprehensive report exploring the possible interplay of AI diffusion and political partisanship. The “analysis reveals a strong relationship between AI involvement and local political partisanship. Specifically, our analysis reveals a strong correlation […] between a county’s AI automation exposure score and its Democratic vote in the 2024 election.”

The “report adds to this limited understanding by analyzing the political behavior of voters in the locations where jobs are becoming involved with AI” (➚Brookings Institute)

Regulatory bits

EU regulators have accused Apple of falsely blaming EU tech rules for its decision not to roll ​out its upgraded assistant Siri AI in the European Union for now (➚Reuters). The EU commission stated that “The decision not to roll out Siri AI in the EU is Apple’s and Apple’s only”. A spokesperson further said that Apple “instead of trying to find a suitable compliance solution, Apple simply made a request to the European Commission to be exempted from their interoperability obligations under the DMA — and this for at least 18 months. That’s not an option.”

It has become rather common for US big tech companies to blame EU regulation for delays when rolling out new products, likely to use their user base to put pressure on regulators.

The German government has announced plans to set up an AI Security Institute: “Furthermore, the National Security Council assessed the consequences of advanced AI models for cybersecurity in Germany and decided to establish an AI security institute. This institute is intended to pool our capacities for the analysis of the capabilities of modern AI models including their risks, to intensify the exchange of information with comparable foreign institutes and to work towards uniform standards in dealing with AI with international partners.” (➚German Government Announcement [German])

Data centres: New lawsuit and polling

A new lawsuit has been filed against SpaceX’s datacentre in Memphis: Three residents of Southaven, Mississippi, have filed a complaint claiming that the data centre is a “public nuisance through excessive ​and offensive noise”. (➚Complaint 3:26-cv-00148-MPM-RP) They are asking the court to certify the lawsuit for class action, and eventually for “compensatory, statutory, and consequential damages, along with punitive, treble or other multiple damage”.

Meanwhile, a poll conducted by “Public First” found that people in the US had the lowest appetite for building computing power among the 15 large countries. Just 26% of Americans supported increased construction of data centres. Trump voters were more likely to support data centre construction, but only at 37% (➚🔒Financial Times).

The majority of new AI data centres in the US are to be built on drought-hit land, an analysis by The Guardian finds: The facilities are to be built in some of the driest areas (➚The Guardian): “Of 809 planned datacenters, 517 are in locations that have been in drought conditions throughout the past year, according to data from Cleanview and the federal government”.

Legal bit

A District Court in Munich, Germany, has prohibited Google from disseminating untrue factual assertions about two Munich publishers in its “Overview with AI”. The AI had incorrectly assigned information of other companies to the plaintiffs’ company.

The judgement is significant because the court says that from the point of view of average users, the AI summary acts like direct information from Google — not like a mere passing through of foreign content.

For this reason the court ruled that the limited liability of search engines for third-party content was not applicable to the “AI Overview” (➚Heise Online [German]).

UN Dialogue

The UN has announced its “AI Dialogues” to “ensure that governance reflects the priorities of all nations, not just the most technologically advanced and that the benefits of AI are shared by all”, and opened the events for registration.

It aims to “share best practices and lessons learned, and facilitate open, transparent and inclusive discussions on artificial intelligence governance”. The events will be livestreamed on 6th and 7th of July. (➚UN Global Dialogue)

MIT AI Risk Initiative

MIT has released the report of its “AI Risk Initiative”. The report reveals how 272 experts assess the severity of AI risks across various sectors and how to mitigate them (➚MIT Paper). It finds that “18 of 24 risks have more than a 10% chance of causing catastrophic outcomes”. (➚MIT AI Risk Initiative)

Copyrighted bits

While the massive settlement in the class action lawsuit by writers against Anthropic is still not approved by the court (➚Reuters), we would like to recommend Mishcon de Reya’s tracker of copyright lawsuits against the AI industry (➚Mishcon De Reya): The tracker “an insight on the various intellectual property cases relating to generative AI going through the courts, as well as anticipated policy and legislative developments”.

Science Bits

A new paper published in Springer Nature by Lin et al. attempts to look at instances of AI “that shape social schemas through subtler manners”. The paper argues that “many existing AI systems should be identified as what Liao and Huebner called oppressive things when they function to manifest oppressive normality”.

“AI systems can perpetuate oppressive normality through the content, namely, the explicit output it presents. […] for instance, they associate high-paying occupations with predominantly images of men with lighter skin tones and associate low-paying occupations with predominantly images of women or people of color”.

Lin, Ta., Huang, L.TL. AI, Normality, and Oppressive Things. Minds & Machines36, 26 (2026).

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.