Mediaeater Digest April 5th, 2025

DOGE Is Planning a Hackathon at the IRS. It Wants Easier Access to Taxpayer Data https://www.wired.com/story/doge-hackathon-irs-data-palantir/ AI giants reject government’s approach to solving copyright row https://www.thetimes.com/uk/technology-uk/article/google-openai-reject-copyright-plan-bnnzztts9 The Llama 4 herd: native multimodal AI https://ai.meta.com/blog/llama-4-multimodal-intelligence/  (weekend drop) ai-2027.com https://ai-2027.com/ TV Garden  | Radio Garden Organising music for organic discovery https://www.bbc.co.uk/rd/articles/2025-03-orbit-music-discovery-organise-analysis/ On the foolishness of “natural language programming” https://www.cs.utexas.edu/~EWD/transcriptions/EWD06xx/EWD667.html Study finds solo music listening boosts social well-being https://phys.org/news/2025-03-solo-music-boosts-social.html Deepsite - (just mind blowing) https://huggingface.co/spaces/enzostvs/deepsite The case against conversational interfaces https://julian.digital/2025/03/27/the-case-against-conversational-interfaces/ ERA Protocol - “Reflexive modeling is not trust. Flattery is not alignment. Containment is not cooperation.” https://github.com/parkertoddbrooks/ERA-Protocol/tree/main  (this is critical) Globalizing Productions with Netflix’s Media Production Suite  (open source for scale) https://netflixtechblog.com/globalizing-productions-with-netflixs-media-production-suite-fc3c108c0a22 United States Disappeared Tracker  (map of the destroyed the ideal of America) https://public.tableau.com/app/profile/danielleharlow/viz/UnitedStatesDisappearedTracker/Map The White House Correspondents Association Speaks Cowardice to Power (Joining Columbia University and the law firm Paul, Weiss  in fealty) https://daringfireball.net/linked/2025/03/31/whca-speaks-cowardice-to-power An image of an archeologist adventurer who wears a hat and uses a bullwhip https://theaiunderwriter.substack.com/p/an-image-of-an-archeologist-adventurer Hardware:screens https://usetrmnl.com/ Severance — Music To Refine To feat. ODESZA | Apple TV+  (8hrs+ hrs)  https://www.youtube.com/watch?v=JRnDYB28bL8 Sec-Gemini v1, a new experimental AI model focused on advancing cybersecurity AI frontiers. https://security.googleblog.com/2025/04/google-launches-sec-gemini-v1-new.html Self-Fulfilling Misalignment Data Might Be Poisoning Our AI Models https://turntrout.com/self-fulfilling-misalignment