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. Feel free to subscribe.
As usual, we will mark all external links with “➚” and all potentially paywalled links with “🔒”.
Why is the engineering problem of AI safety perceived in public discourse as a breach of morality, and why is it dominated by the frames of containment or misalignment, asks Ioannis Akingontein “The Humanization of Artificial Intelligence”.
We are told that AI is inevitable and will affect all aspects of our lives. And now that, for the first time, an employee has been granted an exemption from using AI under the Civil Rights Act, we wonder if “Avoiding AI could become a constitutional right”.
Google, in a classic “don’t be evil” style, is now telling schools they will lose student access to Google’s search engine if they block its AI tools. After upsetting some school leaders by pushing its AI tools into schools, Google now gives them the option to restrict the tools, but says that if school leaders want continued student access to search within the company’s ecosystem, they will be forced to embrace Google’s A.I. In a statement Google called this “evolving our approach” (➚🔒New York Times).
In the still ongoing NAACP v SpaceX lawsuit over SpaceX’s data centre in Memphis, Judge Debra M Brown has ruled that SpaceX’s gas turbines are not — as SpaceX had claimed — a “mobile source”, and therefore are not exempt from permits.
She also found that a citizens lawsuit does not violate Article II of the constitution (SpaceX, and the intervening Trump administration, had claimed it infringes on presidential powers). For those reasons SpaceX’s motion to dismiss the case is denied, and the case will proceed (➚Case: 3:26-cv-00074-DMB-JMV).
Talking about AI and the environment, the BBC has published a deep dive into the question of what the UK’s data centre plans will do to its climate goals (➚BBC).
Meanwhile, in the Netherlands, as part of two days of protest against AI and data centres, dozens of protesters have broken into and occupied a Microsoft data centre construction site in Amsterdam (➚Reuters).
On the more absurd side of the discussion, OpenAI is reportedly planning to rename its data centres to “compute campuses” (➚🔒The Verge).
Meta has settled claims with a second plaintiff in a prominent social media addiction case (➚NBC). In the weeks leading up to trial, TikTok and YouTube reached agreements to resolve the claims. Meta had repeatedly denied that its apps are purposefully harmful and addictive for young users.
Meanwhile, the London’s High Court has dismissed Meta, TikTok, and X’s challenge to Ofcom’s Online Safety Act information requests (➚Reuters). The case is one of several brought over parts of the 2023 act setting tougher standards for social media platforms.
The UK government has reportedly decided to extend Palantir’s controversial £330m health contract while “drawing up plans” to replace it with a British alternative. However, it is expected to take two years to do so (➚🔒Financial Times).
Considering that many parts of the NHS don’t even use the system, and its effectiveness has been debunked again and again, this is strange decision and there was more than enough time to consider alternatives.
The European Commission has announced it will develop age-appropriate reference learning guidance for AI literacy, based on the EU-OECD AI Literacy Framework, and pilot an AI Literacy Certificate for learners from the school year 2027/28. This will be part of its Education and Skills package “to tackle declining basic skills, rapid digital change and teacher shortages” (➚EU Press Release).
The biggest names in AI are backing a new way to opt out of scraping (➚🔒The Verge): OpenAI, Google, Microsoft, and Anthropic say they will adopt a new standard that publishers can use to opt out of being scraped by AI. The shared “policy token” signal, is essentially an extension of the robot.txt protocol.
A new preprint paper examines how AI chatbots shape the relational conditions of emotion coping. The authors argue “that responsible AI should respond to individuals’ situated relational conditions rather than provide general-purpose support”.
Liu, J., Ishfaq, N. (2026). “I Talked an AI Chatbot, So What’s Next?” How U.S. Young Adults Imagine Responsible AI for Emotion Coping. arXiv preprint arXiv:2609.25311.
Another preprint paper from September argues that the use of AI in military operations “produces a concrete accountability gap”.
“The ethical challenge posed by these systems is fundamentally epistemic: not just whether autonomous weapons should be permitted to kill, but whether the conditions for responsible human judgment can survive when critical functions are delegated to opaque algorithms.”
It concludes that “effective governance of military AI requires not only technical constraints but the institutional infrastructure to keep human judgment meaningful”.
Drapier, N., Mauberger, F., Chetouani, A., Chateigner, A. (2026). The Ethics of Artificial Intelligence in Military Operations. arXiv preprint arXiv:2609.26507.
(We looked at questions around accountability and AI in autonomous weapon systems last month in “AI, Autonomous Warfare and the Severance from Responsibility” in Misaligned.)
If you stumbled across this article on the web, subscribe to the “misaligned bits” newsletter and follow Misaligned on Medium, LinkedIn, Threads or Mastodon.
