On January 16, OpenAI announced the launch of a low-cost subscription service called "ChatGPT Go" in a blog post, which costs $8 per month. The service has been available in 171 countries since August of last year, aiming to provide more users with access to AI. OpenAI also stated that it plans to test advertisements for free and Go versions in the United States, while establishing a set of principles to protect user privacy and ensure the independence of ads from model responses.

According to "Information," OpenAI expects consumer subscription users to reach 122 million this year. If this number is achieved, it would significantly change the scale of user interaction with this consumer-level AI product. A large user base often leads to potential security issues, increasing latency and cost pressure on reasoning workloads. This makes the team need to adopt measures such as caching, batching, and secure prompts more proactively to address possible challenges.

At the same time, OpenAI's public principles emphasize not selling user conversation data to advertisers and are committed to protecting user conversation privacy. These measures aim to address common privacy and monetization concerns when ads are combined with conversational AI.

Observers should pay attention to the scope and form of OpenAI's first ad testing in the United States, as well as metrics related to opting out of ad personalization. Additionally, any third-party audit or red team report will also be a key focus. For the engineering team, it is important to note whether OpenAI will make adjustments in rate limiting, communication of service level agreements (SLAs), and new telemetry tools to distinguish advertising service behavior from model response behavior.