A pressing question that comes into mind is what this means for users who didn't sign up to be a part of this ad ecosystem. OpenAI's US privacy policy (updated on April 30) now formally states that it receives purchase data back from advertisers and shares limited user identifiers outside marketing partners. The language, as expected, is careful, but the direction certainly isn't very clear. This should matter for users, as most users are unlikely to read a privacy policy and its subsequent updates. A significant and growing number of young adults, and even the elderly are turning to AI platfo
I didn't realize ChatGPT was serving ads until I dug into how it actually worksIf OpenAI's new direction bothers you as a privacy-conscious user, you may have more options than you might think. Local AI models have come a long way, and are now a viable alternative to cloud-based platforms. Since these models run entirely on your own hardware, no information is sent to a server, and have no advertising business model to answer to. For those who'd rather stay within the OpenAI ecosystem, upgrading to a paid tier removes ads entirely. It does feel worth being clear about what that decision involves, because, after all, OpenAI designed the free tier knowing that the ad exper
I didn't realize ChatGPT was serving ads until I dug into how it actually works… The framework is called Constitutional AI CAI . It is a novel training approach pioneered by Anthropic that uses a written set of principles to help the model self-critique and improve its own outputs. …
… The framework is called Constitutional AI CAI . It is a novel training approach pioneered by Anthropic that uses a written set of principles to help the model self-critique and improve its own outputs. …
… By using an agentic layer, I can let the model actually go out and find information or perform tasks rather than just guessing based on its training data. …
… The framework is called Constitutional AI CAI . It is a novel training approach pioneered by Anthropic that uses a written set of principles to help the model self-critique and improve its own outputs. …
… Local AI models have come a long way , and are now a viable alternative to cloud-based platforms. Since these models run entirely on your own hardware , no information is sent to a server, and have no advertising business model to answer to. …
… Model quality, pricing, reasoning effort, cache behavior, training defaults, quotas, and availability are all controlled server-side, and the past year has shown how many of those dials a cloud vendor can turn in either direction. …
… That said, Google finally launched a model that pulled me back in and it's actually worth running. Google launched four new open-source models You don't need a home lab to use these models A few weeks ago, Google launched its newest family of open-source AI models : Gemma 4. …
… The training data situation isn't exactly clean if I'm being honest - Google got sued back in 2024 over Imagen being trained on the LAION-400M dataset, and YouTube videos are apparently being used to train Veo too. …
… The risks of a scratch disk You might wipe your entire operating system For creative professionals, you'll find that when video editing 8K files or even AI model training, you need a scratch disk, which is a place for temporary high-speed data. …
… Being able to upload research papers, drafts, transcripts, and reference docs into one space and have the AI work across all of them is exactly the kind of AI application I need in my life. In this case though, it isn't really the model doing the heavy lifting. …