After its initial launch exclusively for ChatGPT Plus subscribers in July, OpenAI has now made the custom instructions feature available to all users. This enhancement introduces a new feature in the ChatGPT profile settings, accessible via desktop or iOS platforms, and applies the chosen preferences across all conversations.
Instead of needing to provide specific instructions at the start of each new chat, the system now adjusts its responses automatically based on the user-defined descriptions in the settings. By completing the fields within ChatGPT settings, users can observe the system’s adaptive behavior.
Within the custom instructions section of the settings, there are two text entry fields. The first inquire about the user’s desired information for ChatGPT to be aware of. It even offers suggestions like “Where are you based?” and “What do you do for work?” The second field seeks guidance on how ChatGPT should respond, presenting questions such as “How formal or casual should ChatGPT be?” and “Should ChatGPT have opinions on topics or remain neutral?”
Do you know that Google has taken a huge step in enhancing the security of its Chrome browser by halving the duration for releasing critical security updates? Previously, Google released security upgrades for Chrome every two weeks. However, in light of the growing threat posed by hackers and other malicious actors, the company has opted for a more frequent patch release cadence:
This open-ended feature holds immense potential, leading to a variety of intriguing applications. Some users have explored unconventional uses, while others have focused on practical adjustments. Here are some examples of how people are leveraging this novel option:
One user shared a ChatGPT text output example, claiming that by utilizing a specific setting involving 1,500 repetitions of the letter “a” and an incomplete sentence (a known token repetition hack), the system generated content that seemed to touch upon prohibited topics like unprompted self-harm, nudity, and fictitious personal details.
Many are discovering that custom instructions contribute to enhanced accuracy and effectiveness in ChatGPT’s responses. Amidst these scenarios, a unique literary challenge has emerged as a custom instruction.
OpenAI emphasizes its commitment to not employ instructions that breach its usage policies. Concerning privacy, OpenAI acknowledges the possibility of utilizing settings that inadvertently disclose personal information unless users choose to refrain from saving their chat history.
Although a collection of more general presets with simple yes-or-no options—such as “orient answers toward kids” or “assume user has AI expertise”—might have felt less intrusive than requesting users to articulate personal details, OpenAI has chosen this path for introducing customization. Thus, users are encouraged to safeguard their data as they embrace the era of ChatGPT’s custom instructions.