Resistance meets in London, Palantir again, the AI industry has opinions on regulation, people have opinions on AI and LLMs can quietly steer your opinion.

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 “🔒”.

Misaligned Recap

Last week the “AI Resist List” came together in London at King’s College to present the various projects involved in the network. We wrote about the AI Resist List, a resource for resistance against the Ai industry, before.​

British bits

A second parliamentary committee has now urged Labour to scrap Palantir’s £330m contract with the NHS that will come up for renegotiation next year, increasing pressure on the incoming prime minister over government deals with the US tech company. MPs on the health and social care select committee want the NHS to cut ties with Palantir and the NHS to find a replacement (➚The Guardian).

The U.K. AI Security Institute, formerly known as the AI Safety Institute, has reportedly folded its societal resilience team that researched a broad range of AI risks, spanning fraud to child sexual abuse, into the team that researches how individuals might be affected by AI tools. Two people who spoke to Politico said this “happened in line with the loss of almost all the societal resilience team’s researchers” (➚🔒Politico).

Meanwhile, the government’s hyperscaler darling Nscale has been told its 90MW grid connection will not be ready in time for its Essex data centre’s 2027 opening. Which might sound worse than it is considering that Nscale, which has raised billions without having built a data centre, has not even started building anything there (➚Data Centre Magazine).

Data Centres

While we are on the topic of AI data centres: While the lawsuit against running gas turbines without permissions at SpaceX’s data centre in Memphis is ongoing, Elon Musk has quietly bought a gas turbine company for a $1 billion (➚Electrek). Which rather indicates that the whole gas turbine thing is not as temporary as SpaceX claims, or they do not care about emissions whatsoever.

The White House is planning to ‌bring together utility companies and data centre developers to make a “voluntary pledge” designed to ensure that rapid growth in electricity demand from artificial intelligence does not drive up power bills for households and businesses, thereby for the first time acknowledging that there is a problem here (➚Reuters).

Microsoft, Amazon and Google’s collective carbon emissions in France have increased by nearly a fifth in the past year, driven largely by data centre construction: Over the last 12 months ending March 2026, the three tech companies emitted 119m mTCO₂e (tonnes of carbon dioxide equivalent) (➚The Guardian).

People’s opinions on AI

Over 40% of people are limiting AI use, as popularity starts to wane: People are increasingly choosing to limit or reject the use of AI tools, according to a new report published by the Digital Futures Institute at King’s College London. Between October 2023 and June 2026, the share of UK adults who think AI has more risks than benefits has risen from 48% to 52% (➚King’s Digital Futures Institute).

Those who see more benefits than risks has fallen from 38% to 34%. The most common reasons for limiting use is concern about data privacy, security and compliance (29%), and a preference to keep doing work the way it is currently done at 22%. The trend shows that AI is still becoming becoming more unpopular.

Meanwhile, the AI Policy Institute found that US voters of both parties want tighter AI regulation. In addition, 47% percent of respondents said they would allow data centres if the AI systems being developed had safety and security standards, while 38% said they would ban data centres entirely. (➚AI Policy Center)

Not hiring, just firing

Dozens of Meta employees have filed a lawsuit against the company claiming that Meta used AI tools to tag workers for mass layoffs. The employees allege that those AI tools targeted them after they asked for maternity leave or disability accommodation. The lawsuit further alleges that the company’s AI tools gather data on employees’ performance rankings, productivity and other metrics, inputs that do not exist when workers are on medical or family leave. (➚The Guardian).

Talking about Meta: After announcing that it would allow users of its Meta AI chatbotto tag public-facing accounts on Instagram and re-use the content of those accounts to create new AI-generated material, Meta had to backtrack the plans after the not unexpected backlash (➚BBC).

Regulatory bits

The Dutch government has released a paper on the Netherlands’ International AI Strategy and for once we have a government strategy paper that at least mentions human rights and regulation: “The Netherlands will actively contribute to the further development and effective implementation of the AI Act by strengthening European coordination and practical exchange” (➚Netherlands’ International AI Strategy).

Industry opinions

Sam Altman, CEO of OpenAI, which has spent millions lobbying against the EU Ai Act, has written an opinion piece in the Financial Times calling for a “US-led” global framework that guarantees the safety of AI (➚🔒Financial Times). One day later, Google’s Demis Hassabis also called for a global “US-led” AI watchdog (➚🔒Financial Times). A mere coincidence and in no way coordinated, we must assume.

Other bits to read

Tech Policy Press has published an article on how is India is building “surveillance infrastructure on broken data and bad policing”, including “police standing orders explicitly allow for surveillance of entire families based on their association with ‘habitual offenders’”. (➚Tech Policy Press)

The British Ada Lovelace Institute has published an article on the “UN Global Dialogue on AI”. It concludes that “with political will, funder investment and an institutional infrastructure that goes beyond listening, the UN Global Dialogue can connect to the wealth of community assemblies and help pave the way to participatory AI governance.” (➚Ada Lovelace Institute)

Science bits

A study at the University of Oxford finds that LLMs consistently changed the direction of social media posts on contested topics, such as climate change or abortion, even when explicitly instructed to preserve the original meaning. (➚University of Oxford)

The researchers LLMs from different providers to transform human-written texts on contested topics into improved social media posts. They then analysed whether the AI-generated versions systematically changed the position expressed in the original posts.

The researchers conclude that “that biases introduced by AI in human-to-human communication can be amplified through a social network and shift collective opinion” and “evidence of a directional bias in favor of the pro-life stance on abortion-related posts, and traced this to one specific prompt components”.

Tsirtsis, S., Rawal, K., Russell, C., Mittelstadt, B., Wachter, S. (2026). AI-Mediated Communication Can Steer Collective Opinion. arXiv preprint arXiv:2605.16245

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.