Oversight Board Finds AI Chatbots Echo Authoritarian Speech Limits as Thinking Machines Debuts Inkling
Major AI systems are more likely to refuse to criticize restrictive leaders and governments, according to a Meta Oversight Board study reported by the Inquirer. Testing 10 commercial large language models from companies including Meta, Anthropic, and OpenAI, the study found Claude would make a pamphlet critical of President Trump or King Charles III but declined the same request for Thailand's king, Saudi Arabia's crown prince, or China's leader, an asymmetry the board warns could have "the practical effect of extending the long arm of restrictive governments across borders to limit speech in free countries."
Startup Thinking Machines launched its Inkling model to personalize artificial intelligence, according to a Mix Vale report that could not be retrieved in full. Meanwhile Gen Z is pushing back against the technology outright, per a Channel NewsAsia commentary that cites graduates booing pro-AI commencement speakers and a Gallup study finding the generation is "not convinced that AI enhances creativity or critical thinking," even as Martin Scorsese announced he will embrace AI for storyboarding with Black Forest Labs.
The enterprise story is increasingly organizational rather than technical: HPCwire published an essay by SimOps Foundation Executive Director Wolfgang Gentzsch arguing that the next bottleneck in engineering is no longer technology but organizational capability, the capacity to turn converging HPC, AI, and digital engineering tools into everyday practice. In law, Complete AI Training reports that Landfall IP's managing partners describe their practice as an "AI-native hybrid law firm," with workflows designed around AI from the beginning rather than bolted on. And in mental health, a Talkspace piece distributed by Stacker and carried by News Channel 3-12 traces AI therapy from the 1966 ELIZA chatbot to a 2025 NEJM AI study in which users of the Therabot chatbot saw significant improvements in depression and anxiety symptoms, while clinicians caution AI cannot replace human connection in care.
Sources: