🇷🇺 A robot in a Russian electronics store recently fought back against a customer after being shoved. The footage was wild precisely because nobody could confidently say why it happened. As AI systems become more autonomous and we move closer to AGI, that uncertainty suddenly matters. So far, it has been popular to talk about “AI” as an abstract danger rather than focus on who is responsible when something goes wrong. The liability problem is genuinely difficult. An intelligent system may become complex enough that even its creators cannot reliably predict individual actions, but unpredictability does not automatically answer who should bear responsibility when those actions cause harm. It's a complex problem that needs to be tackled. One that will likely determine not only how we assign responsibility to AI, including whether it has genuine agency, but also shape broader questions about civil liberties and our freedom to create and control technologies capable of causing harm. Writer: Mhedi

Jag vill önska er alla en riktigt glad midsommar. I bästa fall helt utan regn men med sill, tårta och svenska jordgubbar. Och varmt tack till alla er andra som också jobbar idag, ni som ser till att Sverige funkar, så att andra kan vara lediga. Hoppas också ni snart får välförtjänt vila. Själv är jag i Bryssel på EU-toppmöte och försvarar svenska intressen. Det är också ett väldigt fint sätt att fira midsommar.