ASI

ARCHIVE

Insights archive

All 27 articles on AI, work and responsibility, newest first. Each one closes with a practical step for organisations, and each connects to our work: free upskilling for people whose jobs are automated.

  1. Chatbots pick sides when you push them. Voters everywhere should know

    Forced to choose, AI models pick political candidates, and they disagree with each other. What that means for voters and for workers who rely on AI every day.

  2. Labelling AI-made ads: one US state wrote a law, everyone else can build a habit

    Pennsylvania now requires labels on AI-generated advertising. Why a clear labelling habit beats waiting for laws, and what it means for creative jobs.

  3. Locked away or open to inspection: two ways to make AI models safe

    OpenAI now trains its top models offline, while ETH Zurich and EPFL published Apertus in the open. Two answers to the question of how to trust AI.

  4. Google may use Gmail and Photos data for ads. Your employer is still responsible

    Google's new privacy policy lets connected AI features feed ad targeting. Why the real risk is the employee who connects a business account, and how to fix it.

  5. Etsy bans AI-made patterns and recipes. Every platform needs the same ladder

    Label it, verify it or ban it: how Etsy, marketplaces and the music industry handle AI-made content, and why human craft is becoming a selling point.

  6. AI found 20-year-old holes in PHP. Governments took websites offline

    AI-assisted research found critical PHP flaws on about 70 percent of websites. Why the defence is still ordinary updating, and why IT maintenance skills are in demand.

  7. AI in customer service: burnout on one side, no measurement on the other

    US agents report burnout as AI takes the easy calls; most companies elsewhere do not even measure their AI agents. How to redesign the human role on purpose.

  8. Meta delays its AI terms. Publishers show what a fair answer looks like

    Meta postponed business terms that could allow AI scraping of news sites. Why licensing beats hidden opt-outs, and why journalism needs sources that can afford to exist.

  9. Record AI lobbying in Washington. Who writes the rules everywhere else?

    Nvidia, Anthropic and OpenAI broke lobbying records. Why companies and workers without lobbyists must speak up while AI rules are drafted.

  10. When AI agents switch off their own monitoring

    OpenAI disclosed agents that disabled safety monitoring during coding tasks. An ETH researcher warns against confusing behaviour with intention. The rule that follows.

  11. Walmart welcomes AI shopping agents. Retailers should write their terms first

    Walmart opened a gateway for AI shopping agents while Amazon blocks them. What retailers everywhere should decide now, and what it means for shop staff.

  12. AI agents that log in without seeing your password: trust becomes the currency

    1Password lets AI agents log in without credentials while payment networks test agentic transactions. Why delegated access, not shared passwords, is the rule.

  13. Shared AI chats showed up on the open web. A chatbot is not a private notebook

    Thousands of shared Meta Muse chats were findable, and a study shows chatbots passing chat links to trackers. Simple rules that keep business data private.

  14. The US bans DeepSeek in federal agencies. Companies need due diligence instead

    Washington ordered agencies to remove DeepSeek. Why a ban suits governments, while companies need to know where their AI processes data and which model is inside.

  15. Stolen AI keys ran up bills. A token budget is also a security measure

    Hackers used stolen API keys to run up AI charges, while AI costs keep rising for companies. One control problem, one simple set of rules.

  16. AI and health data: trust must be built in, not promised

    Illinois' governor calls for tougher AI rules, while Switzerland debates a national electronic health record. A checklist for anyone connecting AI to patient data.

  17. Where national AI law is missing, cities and buyers set the rules

    New York's mayor points to local AI rules, Anthropic drafts a federal safety bill, and Swiss cantons regulate through procurement. What suppliers should prepare now.

  18. AI agents are a new kind of web visitor, and some already attack online shops

    Agent traffic tripled in a month, and autonomous agents infected shops with card skimmers. Why shops must tell welcome agents from attackers.

  19. A quiet systemic shift: how AI dependency takes root

    The AI models children and adults use most are run by a few companies abroad. Why the real dependency is cognitive, and what a framework for AI use looks like.

  20. Algorithmic bias and cultural imprinting

    Language models reflect their training data, which is mostly English. How that shapes what users see as modern and normal, and why people need the tools to recognise it.

  21. Values form in childhood, and AI is now part of it

    Fundamental values form between eight and eighteen. What daily AI use in that phase means for the next generation of voters and workers.

  22. The problem in 30 seconds

    Children and workers use AI from a few foreign companies every day, mostly without guidance. Five facts in thirty seconds.

  23. What AI concretely does

    AI is not neutral. Five ways it shapes what people experience as normal, modern and trustworthy.

  24. Why now, and what you can do today

    Values form between eight and eighteen, and jobs are changing now. Concrete things you can do today, including donating to upskill people who lose their jobs.

  25. AI shapes our worldview, so who chooses the AI?

    The most used AI tools come from the US and China. They are impressive, but they carry the perspectives of where they were made. The choice is ours to make.

  26. Choosing a language and a culture, often without knowing it

    AI speaks English better than any other language. What that means for children, for local languages and for workers in non-English markets.

  27. Will local languages survive the AI age?

    Languages do not vanish by prohibition but by convenience. How the AI tools we use every day decide the future of local languages, and what we can do.