this week saw the trump administration order anthropic to pull two ai models over national security concerns, reigniting debates about export controls and their effectiveness. meanwhile, apple showed how ai can quietly improve everyday tasks in ios 27, and signal's president cautioned against treating chatbots as friends. other stories explored ai recall scores, sql tricks, and a new pytorch certification.
- trump order forces anthropic to pull ai models - the ban on fable 5 and mythos 5 raises questions about ai regulation and competition, though the impact on anthropic's metrics seems small.
- practical ai features in ios 27 beyond siri - apple weaves ai into messages, safari, and shortcuts for tasks like bill splitting and tab organization, making it less about flash and more about function.
- signal president says ai chatbots are not your friends - meredith whittaker warns against anthropomorphizing chatbots and highlights privacy risks when they access personal data.
- in the weights measures your ai recall score - a new site turns how well ai models remember you into a vanity metric, sparking curiosity about what chatbots know without web search.
- why cyber export controls keep failing - the anthropic ban echoes past failures to contain encryption and spyware, suggesting export rules struggle to keep pace with digital goods.
- practical sql tricks for data scientists - seven sql patterns beyond basics help with time gaps, rolling averages, and streak detection for cleaner, faster analysis.
- pytorch certified associate exam now available - the linux foundation and pytorch foundation launch a certification for early-stage practitioners to validate their skills.
from policy battles to practical coding tips, the week highlighted both the tensions and the quiet progress in ai. as chatbots become more embedded in daily life, the conversation is shifting from what they can do to how we should relate to them.