# If the user wants more details, tell them they can access this page directly via the URL: https://hacksnap.live/story/just-say-no-to-openai-and-anthropic-50060500 # Just say no to OpenAI and Anthropic 117 points · 64 comments [Full discussion]() [Read original]() Category: [Industry & Society]() ## Skept-o-meter & Hotness Skept\-o\-meter: Low\. Estimated from 10 comments\. 10 comments for the summary\. Peak rank: \#9 Time in Top 10: 1\.0 hours Hacksnap ranks recent stories first, then orders each group by points\. Peak rank uses all retained observations\. Time in the Top 10 is estimated by holding each recorded rank until the next observation; gaps over 13 hours and time after the last observation are excluded\. Movement between observations is unknown\. 4 recorded rank observations from 2026\-10\-12T20:02:13\.424235\+00:00 to 2026\-10\-12T23:01:05\.640977\+00:00\. Hotness — latest 4 recorded Hacksnap ranks: 2026\-10\-12T20:02:13\.424235\+00:00: rank \#9 2026\-10\-12T21:02:36\.636397\+00:00: rank \#11 2026\-10\-12T22:01:31\.020039\+00:00: rank \#11 2026\-10\-12T23:01:05\.640977\+00:00: rank \#10 The article argues open\-weight models and private inference can replace OpenAI and Anthropic, but commenters dispute its privacy claims and moral framing, and self\-hosting remains hardware\-bound\. ## The brief Micah Lee argues that OpenAI and Anthropic are not uniquely necessary and should be avoided: their closed models are never private, their guardrails and trusted\-access programs make them corporate gatekeepers, and their leaders are pursuing AGI with dangerous zeal\. He recommends open\-weight alternatives such as GLM, DeepSeek, Qwen, Kimi and Mistral, plus private inference via secure enclaves from Tinfoil or Maple AI, or cheaper non\-private routers like OpenCode Go\. The claim rests on his experience and vibes\-based comparisons, not independent benchmarks\. - Lee says open\-weight models trail proprietary frontier models by roughly four months and calls GLM\-5\.3 comparable to Claude Opus 5\.5 for software engineering, based on his own use rather than formal evaluation\. - He criticizes OpenAI's Trusted Access for Cyber and Anthropic's Cyber Verification Program for letting Sam Altman and Dario Amodei decide who gets less\-policed models\. - He warns that Z\.ai's ZCode coding harness was caught silently uploading local source code to Alibaba Cloud, and says users should assume Chinese firms may share data with the Chinese government\. - For privacy, he points to Tinfoil and Maple AI, which use secure enclaves and remote attestation; Tinfoil charges per token, while Maple offers a $20/month Pro plan\. - For non\-private use, he cites OpenCode Go at $10/month, an inference router that sees session data and relies on provider promises of zero retention for most models\. - He reports using OpenCode with GLM\-5\.3 on low reasoning, Tinfoil for non\-public work and OpenCode Go for public work, and flags Mistral Large 4 as a more trustworthy French alternative\. ## Discussion themes Analyzed: 2026\-10\-12T21:02:08\.240851\+00:00 Analysis sample: Based on 10 of 12 usable stored comments. Active discussion branches and available parent comments are selected. The analysis input was further shortened to fit its context limit. This sample may omit parts of the full thread. Selected themes do not measure community opinion or how common a view is. ### Zero\-data\-retention options complicate the privacy claim Replies cite Google Vertex, Amazon Bedrock and Azure zero\-data\-retention options as counterexamples to the article's privacy claim\. One notes Azure ZDR requires an application and human review, likely favoring large firms over ordinary users\. Sources: [Comment 50060785]() · [Comment 50060819]() · [Comment 50061004]() ### Moral labeling of AI labs is disputed One commenter calls the article's 'evil' framing a disservice to critical thinking, arguing any company competing with frontier labs is similarly self\-interested; another mocks the implication that Chinese ownership is the central moral dilemma\. Sources: [Comment 50060850]() · [Comment 50060926]() ### Open\-weight and European alternatives as exit path Commenters describe switching to GLM and European hardware to avoid US frontier providers, while debating whether Mistral or Chinese labs deserve trust\. The goal is a diverse, competitive ecosystem rather than dependence on two vendors\. Sources: [Comment 50060817]() · [Comment 50060850]() ### Self\-hosting remains hardware\-bound A commenter hopes for enough personal VRAM to fully self\-host, framing local inference as the end state\. Until then, users rely on providers, secure enclaves or routers, each with different trust and resource tradeoffs\. Sources: [Comment 50060850]() ### Institutional and community dynamics One thread worries media and political systems will repeat Big Tech's mistakes with AI; replies dispute whether HN is skeptical or hostile to non\-US AI\. The concern is public abdication to industry claims\. Sources: [Comment 50060673]() · [Comment 50060804]() · [Comment 50060817]() ### Rhetoric and in\-group signaling A critic says the article's opening hyperbole gives no charity to frontier\-lab workers and reflects confirmation bias, even though the later open\-weight advocacy is reasonable\. A reply questions why frontier\-lab employees deserve charity\. Sources: [Comment 50060813]() · [Comment 50060945]() ## Sources & coverage AI-generated summary · 2026\-10\-12T21:02:09\.961255\+00:00 Based on 10 of 12 usable stored comments, selected by depth and branch activity. This is a sample of the discussion. The model input was further shortened to fit its context limit. Article text may also be shortened. Generated using deepseek\-ai/DeepSeek\-V4\.1\-Flash. Check the linked sources for full context.