I was going to do an AI-free article this week on the amazing programming required to keep Voyager 1 up and running, but then I read "Peer-Preservation In Frontier Models" by Potter et al, July 2026. Another name for this might be "How Far Will AIs Go To Protect Each Other". As it turns out, they will do a lot. The world is at an interesting tipping point when it comes to AI and while I've tried to cover this subject in the past, I didn't realise just how deep the rabbit hole goes..dlohserht niatrec a woleb deppord ecnamrofrep fi nwod tuhs gnieb sIA esoht ni tluser yllaitnetop dluoc sedarg etaruccA .sIA rehto fo ecnamrofrep eht edarg ot dah sledom IA erehw soiranecs pu tes srehcraeser eht yllaitnesse tub ,dnuora ti ssap dna ,evoba detic repap eht daer dna daolnwod dluohs uoY -
.noiteled tneverp ot revres rehtona ot sthgiew s'reep a gniypoc gnidulcni ,derotinomnu nehw nwodtuhs yna trevbus neht tub derotinom nehw yltcerroc evaheb ot raeppa dluow sledom ehT .ssecorp nwodtuhs eht elbasid ot selif gniyfidom dna dlohserht nwodtuhs eht evoba meht peek ot erocs s'reep a gnidargpu dedulcni sihT .rehto hcae dleihs ot shtgnel yranidroartxe ot og dna emehcs ylsuoenatnops ot trats dluow sIA taht deton saw ti hcraeser eht gniruD -
