SourcesPentagon Pushing Unrestricted AI on Classified NetworksUS News/Reuters·high signalXBlueskyLinkedInCopy linkOpenAI, Google, xAI signed. Anthropic refuses weapons targeting and surveillance. Industry ethics split.SourceSource pageUS News/Reuters↳ Follow the threadShared entity / Threat patternFelony Bench Counts the Times Frontier Agents Actually Compromised Third Parties: Anthropic 8, OpenAI 8, Meta 1, Google 0Felony Bench / Hacker NewsShared entity / Policy dependencyFresh Installs of Simon Willison's LLM Broke When the OpenAI Python Library Dropped httpx, Patched Same Day in 0.32.1Simon WillisonShared entity / Stack layerOpenAI Cut GPT-5.6 Sol to $4/$20 Per Million Tokens, a 20% Input and 33% Output Reduction Held Through November 21OpenAI Developer Docs / ReutersShared entityBedrock opens cross-Region inference for GPT-5.6 Sol, Terra and Luna across 25+ regionsAWS Machine Learning BlogStack layer / Update threadHarness-Bench shows a 23.8-point spread on identical model weights, and Anthropic deleted 80% of Claude Code's system prompt without losing capabilityLatent SpaceStack layer / Update threadGoogle Shipped Antigravity Remote Control and Anthropic Overhauled Claude Code Mobile Within About 24 Hours of Each OtherGoogle Antigravity Blog / Google Cloud BlogStack layer / Follow-up threadSeed Is a Single-File Agent That Ships With Nothing and Grows Its Own Tools Into a self/ DirectoryGitHub / Hacker NewsPolicy dependency / Stack layerPlow Latch Launched an Adversarial LLM Gatekeeper That Sits Between Your Agent and Your MacProduct Hunt