-
Join the MyBroadband community
South Africa’s biggest forum. Discuss, discover, and connect with thousands of members.
-
Question of the Day: If you could pick one South African to have dinner with, who would it be?
ChatGPT
- Thread starter animal531
- Start date
Herr der Verboten
Honorary Master
all llms are like this some just more than others.
quovadis
Honorary Master
On a similar topic... if it hasn't been posted anywhere... It's kinda funny and somewhat expected if you've used a wide variety of the commercial models...
EMERGENCE WORLD: A Laboratory for Evaluating Long-horizon Agent Autonomy — Emergence AI
Most evaluations of AI agents look like exams: a discrete task, a clean environment, a score in minutes or hours. Emergence World is built for the opposite question—what happens when you let agents run continuously, in a shared environment with real-world signals, for weeks. It is a research platfor
www.emergence.ai
OnlyOneKenobi
Executive Member
- Joined
- Dec 6, 2013
- Messages
- 5,511
- Reaction score
- 8,366
It's more sensitive lately, especially with regards to violence, gore, horror, etc. It's always been sensitive around sex and nudity or even skimpy clothing. With regard to copyrighted characters, it's all over the place.
Herr der Verboten
Honorary Master
…gets even funnier if you own the stack or are looking at how to write the stack and llm tell you its not allowed to assist with your request because your trying to get it to make malware, hacking, penetration, etc - and it was some as simple as middleware auth handler and or sanitizer.
FiestaST
Honorary Master
Lawnchairman
Expert Member
FiestaST
Honorary Master
FiestaST
Honorary Master
Yes — if the forum was referring to this version of ChatGPT, then “Luna” is accurate.
I’m GPT-5.6 Luna. So if someone says “ChatGPT’s name is Luna,” they’re essentially referring to the model/persona name for this version.
Small distinction: ChatGPT is the product; GPT-5.6 Luna is the model you’re talking to.
FiestaST
Honorary Master
FiestaST
Honorary Master
OpenAI is adding visual ads to ChatGPT
OpenAI introduced ads for free and Go users of ChatGPT back in February. Those were pretty simple ads that appeared between chat messages. Now the company is adding a new ad format, which it calls visual ads.
These feature larger images – either several thumbnails or one screen-width image. These aren’t live yet as OpenAI plans to introduce them later this month. This will start with a testing period where visual ads will only be shown when the user generates images with ChatGPT.
However, the visual ads will be presented separately from the generated images and will be clearly marked as an ad. Also, OpenAI says that the ads will have no influence over the AI’s response.
Additionally, the company has put guardrails in place to prevent ads from showing up in chats with “emotionally vulnerable, sensitive or otherwise unsuitable contexts.” There will be continuous monitoring to ensure that these guardrails work, including both automated checks and human oversight.
The press release makes no mention of it, but presumably only free and Go tier ChatGPT users will be seeing the new ads - just like the initial ad rollout.
OpenAI has worked on integrating its ad solutions with traditional advertising networks to make it easier for clients to work with. It has also built new tools to help advertisers measure the impact of their ads – if you are interested, you can read more in this OpenAI post.
www.gsmarena.com
OpenAI introduced ads for free and Go users of ChatGPT back in February. Those were pretty simple ads that appeared between chat messages. Now the company is adding a new ad format, which it calls visual ads.
These feature larger images – either several thumbnails or one screen-width image. These aren’t live yet as OpenAI plans to introduce them later this month. This will start with a testing period where visual ads will only be shown when the user generates images with ChatGPT.
However, the visual ads will be presented separately from the generated images and will be clearly marked as an ad. Also, OpenAI says that the ads will have no influence over the AI’s response.
Additionally, the company has put guardrails in place to prevent ads from showing up in chats with “emotionally vulnerable, sensitive or otherwise unsuitable contexts.” There will be continuous monitoring to ensure that these guardrails work, including both automated checks and human oversight.
The press release makes no mention of it, but presumably only free and Go tier ChatGPT users will be seeing the new ads - just like the initial ad rollout.
OpenAI has worked on integrating its ad solutions with traditional advertising networks to make it easier for clients to work with. It has also built new tools to help advertisers measure the impact of their ads – if you are interested, you can read more in this OpenAI post.
OpenAI is adding visual ads to ChatGPT
These feature larger images than the mostly text ad format that was introduced in February. OpenAI introduced ads for free and Go users of ChatGPT back in...