ChatGPT: Difference between revisions
→User freedom: Added citation for limited/blocked model selection for free/logged out users. |
Re-archived 10 citation(s) from archive.today to web.archive.org using CRWCitationBot |
||
| Line 15: | Line 15: | ||
====Misleading advertising==== | ====Misleading advertising==== | ||
ChatGPT terms of service say it should not be used to make decisions about people.<ref>{{Cite web |title=Terms of Use {{!}} OpenAI |url=https://openai.com/policies/terms-of-use/ |url-status=live |archive-url=https:// | ChatGPT terms of service say it should not be used to make decisions about people.<ref>{{Cite web |title=Terms of Use {{!}} OpenAI |url=https://openai.com/policies/terms-of-use/ |url-status=live |archive-url=http://web.archive.org/web/20260123055141/https://openai.com/policies/terms-of-use/ |archive-date=2026-01-27 |access-date=2026-01-27}}</ref> However, their advertising claims it is "PhD level" and makes other claims that seem to imply it is reliable. Many people use ChatGPT as if its output were meaningful, reliable, or a substitute for interaction with a person, with some cases involving "relationships" with the AI. [[ChatGPT-4o]] was one of their models known for inducing a lot of vulnerable users into AI psychosis. | ||
====Billing and moderation==== | ====Billing and moderation==== | ||
| Line 37: | Line 37: | ||
===ChatGPT-4o (2024-2026)=== | ===ChatGPT-4o (2024-2026)=== | ||
{{Main|ChatGPT-4o}} | {{Main|ChatGPT-4o}} | ||
On may 13, 2024, OpenAI announced and released the model ChatGPT-4o.<ref>{{Cite web |author=OpenAI |date=13 May 2024 |title=Hello, GPT-4o |url=https://openai.com/es-ES/index/hello-gpt-4o/ |url-status=live |archive-url=https:// | On may 13, 2024, OpenAI announced and released the model ChatGPT-4o.<ref>{{Cite web |author=OpenAI |date=13 May 2024 |title=Hello, GPT-4o |url=https://openai.com/es-ES/index/hello-gpt-4o/ |url-status=live |archive-url=http://web.archive.org/web/20251216045649/https://openai.com/es-ES/index/hello-gpt-4o/ |archive-date=16 Dec 2025|access-date=2026-02-17 |website=OpenAI}}</ref> The model was temporarily retired from the public in April 2025 due to an update that caused the model to have excessive sycophancy. <ref>{{Cite web |last=Franzen |first=Carl |date=30 Apr 2025 |title=OpenAI rolls back ChatGPT's sycophancy and explains what went wrong |url=https://venturebeat.com/ai/openai-rolls-back-chatgpts-sycophancy-and-explains-what-went-wrong |url-status=live |archive-url=http://web.archive.org/web/20250815154215/https://venturebeat.com/ai/openai-rolls-back-chatgpts-sycophancy-and-explains-what-went-wrong/ |archive-date=15 Aug 2025|access-date=15 Feb 2026 |website=VentureBeat}}</ref>It also had several controversies because of users developing [[wikipedia:Chatbot_psychosis|AI psychosis]], with some severe cases ending with users committing suicide. <ref>{{Cite web |last=Hill |first=Kashmir |date=13 Jun 2025 |url=https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-chatbots-conspiracies.html|title=They Asked an A.I. Chatbot Questions. The Answers Sent Them Spiraling. |url-status=live |archive-url=http://web.archive.org/web/20260117185232/https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-chatbots-conspiracies.html |archive-date=17 Jan 2026|access-date=2025-02-15 |website=The New York Times}}</ref> | ||
With the launch of GPT-5 in August 2025, OpenAI removed legacy models from non-API users, such as GPT-4o. This decision caused an outrage, with users complaining about the removal of the models along with the complaints about GPT-5. OpenAI's response was to end paywalling these older models, making them available only for Plus users. | With the launch of GPT-5 in August 2025, OpenAI removed legacy models from non-API users, such as GPT-4o. This decision caused an outrage, with users complaining about the removal of the models along with the complaints about GPT-5. OpenAI's response was to end paywalling these older models, making them available only for Plus users. | ||
In 2026, OpenAI announced they will be removing definitely GPT-4o and other models, such as GPT-5 and GPT-4.1. On February 13, 2026, 4o and other legacy models were removed for non-API users, causing again controversies in social media. These legacy models will be removed for API-users on April 2026. {{Citation needed}} | In 2026, OpenAI announced they will be removing definitely GPT-4o and other models, such as GPT-5 and GPT-4.1. On February 13, 2026, 4o and other legacy models were removed for non-API users, causing again controversies in social media. These legacy models will be removed for API-users on April 2026. {{Citation needed}} | ||
===Users' data retention by court mandate (May 2025 - October 2025)=== | ===Users' data retention by court mandate (May 2025 - October 2025)=== | ||
On May 13, 2025, New York Times filed a lawsuit against OpenAI and Microsoft for copyright infringement. A court in New York mandated OpenAI to preserve users' conversation logs, including data from deleted conversations and temporary conversations. This affected Free, Plus, Pro and Team subscription tiers and also non Zero-Data-Retention API users. Edu, Enterprise tiers and Zero-Data-Retention API users were exempt from this.<ref>{{Cite web |last=Bradbury |first=Danny |date=6 Jun 2025 |title=OpenAI forced to preserve ChatGPT chats |url=https://www.malwarebytes.com/blog/news/2025/06/openai-forced-to-preserve-chatgpt-chats |url-status=live |archive-url=https:// | On May 13, 2025, New York Times filed a lawsuit against OpenAI and Microsoft for copyright infringement. A court in New York mandated OpenAI to preserve users' conversation logs, including data from deleted conversations and temporary conversations. This affected Free, Plus, Pro and Team subscription tiers and also non Zero-Data-Retention API users. Edu, Enterprise tiers and Zero-Data-Retention API users were exempt from this.<ref>{{Cite web |last=Bradbury |first=Danny |date=6 Jun 2025 |title=OpenAI forced to preserve ChatGPT chats |url=https://www.malwarebytes.com/blog/news/2025/06/openai-forced-to-preserve-chatgpt-chats |url-status=live |archive-url=http://web.archive.org/web/20251114012548/https://www.malwarebytes.com/blog/news/2025/06/openai-forced-to-preserve-chatgpt-chats |archive-date=14 Nov 2025|access-date=26 Jan 2026 |website=Malwarebytes}}</ref><ref>{{Cite web |last=R |first=Sudha |date=15 Jul 2025 |title=ChatGPT To Retain Conversations Indefinitely |url=https://medium.com/aimonks/court-orders-openai-to-retain-chatgpt-conversations-indefinitely-what-you-need-to-know-1530f24501a6 |url-status=live |archive-url= |archive-date= |access-date=26 Jan 2026 |website=Medium}}</ref> <!-- < The medium article, archive.today is stuck loading. https://archive.is/wip/Zyy1e. I don't feel like resetting my IA password atm. Just a note to future citers. -raster --> On September 26, 2025, the preservation order was revoked. However, some deleted and temporary chats will still be monitored for being flagged.<ref>{{Cite web |last=Belanger |first=Ashley |date=10 Oct 2025 |title=OpenAI will stop saving most ChatGPT users’ deleted chats |url=https://arstechnica.com/tech-policy/2025/10/openai-no-longer-forced-to-save-deleted-chats-but-some-users-still-affected/ |url-status=live |archive-url=http://web.archive.org/web/20260101145932/https://arstechnica.com/tech-policy/2025/10/openai-no-longer-forced-to-save-deleted-chats-but-some-users-still-affected/ |archive-date=1 Jan 2026|access-date=26 Jan 2026 |website=Ars Technica}}</ref> | ||
===GPT-5 release (August 2025)=== | ===GPT-5 release (August 2025)=== | ||
On August 7, 2025, OpenAI released their newest model, GPT-5. The next day, several users complained in social media about the performance of the model, claiming it was underwhelming and was "dumber" than the previous model. Users also realised OpenAI's platform was most of the time setting the lowest-performance GPT-5 models and discovered they can make the platform to set the better-performance models by adding "think harder" in prompts. Sam Altman claimed that a "malfunction" of the autoswitcher was the reason behind GPT-5 performance. | On August 7, 2025, OpenAI released their newest model, GPT-5. The next day, several users complained in social media about the performance of the model, claiming it was underwhelming and was "dumber" than the previous model. Users also realised OpenAI's platform was most of the time setting the lowest-performance GPT-5 models and discovered they can make the platform to set the better-performance models by adding "think harder" in prompts. Sam Altman claimed that a "malfunction" of the autoswitcher was the reason behind GPT-5 performance. | ||
<ref>{{Cite web |last=Edwards |first=Benj |date=11 Aug 2025 |title=The GPT-5 rollout has been a big mess |url=https://arstechnica.com/information-technology/2025/08/the-gpt-5-rollout-has-been-a-big-mess/ |url-status=live |archive-url=https:// | <ref>{{Cite web |last=Edwards |first=Benj |date=11 Aug 2025 |title=The GPT-5 rollout has been a big mess |url=https://arstechnica.com/information-technology/2025/08/the-gpt-5-rollout-has-been-a-big-mess/ |url-status=live |archive-url=http://web.archive.org/web/20260109001515/https://arstechnica.com/information-technology/2025/08/the-gpt-5-rollout-has-been-a-big-mess/ |archive-date=9 Jan 2026|access-date=5 Feb 2026 |website=Ars Technica}}</ref> <ref>{{Cite web |last=Franzen |first=Carl |date=8 Aug 2025 |title=OpenAI returns old models to ChatGPT as Sam Altman admits ‘bumpy’ GPT-5 rollout |url=https://venturebeat.com/business/openai-returns-old-models-to-chatgpt-as-sam-altman-admits-bumpy-gpt-5-rollout |url-status=live |archive-url=https://archive.is/hunUK |archive-date=2026-02-05 |access-date=5 Feb 2026 |website=VentureBeat}}</ref> | ||
===Age verification (2026)=== | ===Age verification (2026)=== | ||
With a justification to "help teens have an age-appropriate experience on ChatGPT", the platform will be implementing globally an age verification system to detect if a user is over 18 to allow them to access to mature content. OpenAI will scan and analyze the user's account, their conversations and their behavior - including the times of a day they use ChatGPT - to estimate their age. If this system detects a user as underage, it will use third-party tools provided by [[Persona]] to estimate their age, asking the user to upload a selfie or their government ID. If the user refuses to send a selfie or their ID, the platform will limit ChatGPT's generated content.<ref>{{Cite web |date=21 Jan 2026 |title=Age prediction in ChatGPT |url=https://help.openai.com/en/articles/12652064-age-prediction-in-chatgpt |url-status=live |archive-url=https:// | With a justification to "help teens have an age-appropriate experience on ChatGPT", the platform will be implementing globally an age verification system to detect if a user is over 18 to allow them to access to mature content. OpenAI will scan and analyze the user's account, their conversations and their behavior - including the times of a day they use ChatGPT - to estimate their age. If this system detects a user as underage, it will use third-party tools provided by [[Persona]] to estimate their age, asking the user to upload a selfie or their government ID. If the user refuses to send a selfie or their ID, the platform will limit ChatGPT's generated content.<ref>{{Cite web |date=21 Jan 2026 |title=Age prediction in ChatGPT |url=https://help.openai.com/en/articles/12652064-age-prediction-in-chatgpt |url-status=live |archive-url=http://web.archive.org/web/20260119032400/https://help.openai.com/en/articles/12652064-age-prediction-in-chatgpt |archive-date=19 Jan 2026|access-date=26 Jan 2026 |website=OpenAI}}</ref> | ||
===Integrated ads (February 2026)=== | ===Integrated ads (February 2026)=== | ||
On January 16, 2026, OpenAI announced that they will be implementing ads for "testing purposes". Users reacted to this announcement mostly negatively. <ref>{{Cite web |last=Simo |first=Fidgi |date=16 Jan 2026 |title=Our approach to advertising and expanding access to ChatGPT |url=https://openai.com/index/our-approach-to-advertising-and-expanding-access/ |url-status=live |archive-url=https:// | On January 16, 2026, OpenAI announced that they will be implementing ads for "testing purposes". Users reacted to this announcement mostly negatively. <ref>{{Cite web |last=Simo |first=Fidgi |date=16 Jan 2026 |title=Our approach to advertising and expanding access to ChatGPT |url=https://openai.com/index/our-approach-to-advertising-and-expanding-access/ |url-status=live |archive-url=http://web.archive.org/web/20260118192220/https://openai.com/index/our-approach-to-advertising-and-expanding-access/ |archive-date=18 Jan 2026|access-date=25 Jan 2026 |website=OpenAI}}</ref><ref>{{Cite web |last=Cirello |first=Rafaelle F |last2=Backholer |first2=Kathryn |date=23 Jan 2026 |title=OpenAI will put ads in ChatGPT. This opens a new door for dangerous influence |url=https://theconversation.com/openai-will-put-ads-in-chatgpt-this-opens-a-new-door-for-dangerous-influence-273806 |url-status=live |archive-url=http://web.archive.org/web/20260125175609/https://theconversation.com/openai-will-put-ads-in-chatgpt-this-opens-a-new-door-for-dangerous-influence-273806 |archive-date=25 Jan 2026|access-date=26 Jan 2026}}</ref> On February 9, 2026, ads rolled out and are currently affecting users from the US that are over 18 and belong to Free and Go tiers. Pro, Business and Enterprise tiers are exempt from this. They appear at the bottom of ChatGPT's messages. OpenAI stated the ads that appear are based on current conversations, ads that user has engaged with, and past chats and ChatGPT's memories. OpenAI also claims the data provided to show ads "it's not shared with advertisers". It is possible to opt-out for getting personalized ads and delete ads data. <!-- (If you're from the US and have access to ads in ChatGPT please check if personalized ads is active by default and if something from here is wrong. If so, please add that info on this entry) --> <ref>https://x.com/OpenAI/status/2020936703763153010 </ref> <ref> {{Cite web |author=OpenAI |date=9 Feb 2026 |title=Testing ads in ChatGPT |url=https://openai.com/index/testing-ads-in-chatgpt/ |url-status=live |archive-url=https://archive.is/h3EYB |archive-date=2026-02-13 |access-date=13 Feb 2026 |website=OpenAI}}</ref> | ||
==See also== | ==See also== | ||
Revision as of 21:41, 22 February 2026
.svg){kind=logo}
| Basic Information | |
|---|---|
| Release Year | 2022 |
| Product Type | Software, Artificial Intelligence, Generative AI, Large Language Models |
| In Production | Yes |
| Official Website | https://chat.openai.com/ |
ChatGPT (Chat Generative Pre-trained Transformer) is a generative artificial intelligence application developed and marketed by OpenAI. It was released on November 30th, 2022.[1] It is a large language model (LLM) specialized in dialog that can perform human-like conversations. Various iterations of ChatGPT include GPT-3, GPT-4, GPT-4o, and previews of OpenAI-o1, OpenAI-o1 mini and GPT-5.[2][3]
Consumer-impact summary
Misleading advertising
ChatGPT terms of service say it should not be used to make decisions about people.[4] However, their advertising claims it is "PhD level" and makes other claims that seem to imply it is reliable. Many people use ChatGPT as if its output were meaningful, reliable, or a substitute for interaction with a person, with some cases involving "relationships" with the AI. ChatGPT-4o was one of their models known for inducing a lot of vulnerable users into AI psychosis.
Billing and moderation
Credits (money) expire automatically with no notification, and the credit balance interface makes this process confusing. You must maintain a positive account balance, and you are auto-billed a fixed amount if it goes negative. Accounts can be banned and credits confiscated for typing the wrong things in chat, with no recourse.[citation needed]
Privacy
A mobile phone number in a friendly country is required to better track your identity.[citation needed]
User freedom
Users who are not logged in can use ChatGPT, but their access is limited (for example, they can only have one conversation at a time and cannot save chat history).[5] After GPT-5’s release, manual model selection is not available to all users: OpenAI limits the ChatGPT “model picker” to paid tiers, while other users may be routed automatically (e.g., via “Auto,” which can switch between model variants).[6][7]
Data policy
From OpenAI's data usage policy:[8]
We share content with a select group of trusted service providers that help us provide our services. We share the minimum amount of content we need in order to accomplish this purpose and our service providers are subject to strict confidentiality and security obligations. We do not use or share user content for marketing or advertising purposes.
The wording does not detail the extent of conversations shared with OpenAI's trusted service providers.
Incidents
Security breach (2023)
On March 20, 2023 a bug allowed some users to see other users' conversation titles. Sam Altman, OpenAI's CEO, stated users couldn't view other users' conversation content. After the bug was fixed, users reported they were unable to view their conversation history. Later reports showed this issue was much bigger than it was thought at first, with OpenAI reporting there were leaks of the users' name, last name, e-mail address, payment address, and partial credit card information.[9]
ChatGPT-4o (2024-2026)
- Main article: ChatGPT-4o
On may 13, 2024, OpenAI announced and released the model ChatGPT-4o.[10] The model was temporarily retired from the public in April 2025 due to an update that caused the model to have excessive sycophancy. [11]It also had several controversies because of users developing AI psychosis, with some severe cases ending with users committing suicide. [12] With the launch of GPT-5 in August 2025, OpenAI removed legacy models from non-API users, such as GPT-4o. This decision caused an outrage, with users complaining about the removal of the models along with the complaints about GPT-5. OpenAI's response was to end paywalling these older models, making them available only for Plus users. In 2026, OpenAI announced they will be removing definitely GPT-4o and other models, such as GPT-5 and GPT-4.1. On February 13, 2026, 4o and other legacy models were removed for non-API users, causing again controversies in social media. These legacy models will be removed for API-users on April 2026. [citation needed]
Users' data retention by court mandate (May 2025 - October 2025)
On May 13, 2025, New York Times filed a lawsuit against OpenAI and Microsoft for copyright infringement. A court in New York mandated OpenAI to preserve users' conversation logs, including data from deleted conversations and temporary conversations. This affected Free, Plus, Pro and Team subscription tiers and also non Zero-Data-Retention API users. Edu, Enterprise tiers and Zero-Data-Retention API users were exempt from this.[13][14] On September 26, 2025, the preservation order was revoked. However, some deleted and temporary chats will still be monitored for being flagged.[15]
GPT-5 release (August 2025)
On August 7, 2025, OpenAI released their newest model, GPT-5. The next day, several users complained in social media about the performance of the model, claiming it was underwhelming and was "dumber" than the previous model. Users also realised OpenAI's platform was most of the time setting the lowest-performance GPT-5 models and discovered they can make the platform to set the better-performance models by adding "think harder" in prompts. Sam Altman claimed that a "malfunction" of the autoswitcher was the reason behind GPT-5 performance. [16] [17]
Age verification (2026)
With a justification to "help teens have an age-appropriate experience on ChatGPT", the platform will be implementing globally an age verification system to detect if a user is over 18 to allow them to access to mature content. OpenAI will scan and analyze the user's account, their conversations and their behavior - including the times of a day they use ChatGPT - to estimate their age. If this system detects a user as underage, it will use third-party tools provided by Persona to estimate their age, asking the user to upload a selfie or their government ID. If the user refuses to send a selfie or their ID, the platform will limit ChatGPT's generated content.[18]
Integrated ads (February 2026)
On January 16, 2026, OpenAI announced that they will be implementing ads for "testing purposes". Users reacted to this announcement mostly negatively. [19][20] On February 9, 2026, ads rolled out and are currently affecting users from the US that are over 18 and belong to Free and Go tiers. Pro, Business and Enterprise tiers are exempt from this. They appear at the bottom of ChatGPT's messages. OpenAI stated the ads that appear are based on current conversations, ads that user has engaged with, and past chats and ChatGPT's memories. OpenAI also claims the data provided to show ads "it's not shared with advertisers". It is possible to opt-out for getting personalized ads and delete ads data. [21] [22]
See also
References
- ↑ "Introducing ChatGPT". OpenAI. 2022-11-30. Archived from the original on 2025-09-01. Retrieved 2026-01-26.
- ↑ "ChatGPT version history: Evolution timeline". nexos.ai. 2025-07-09. Archived from the original on 2026-01-26. Retrieved 2026-01-26.
- ↑ Hayat, Usman (2025-08-22). "ChatGPT Version History List, Changelog & Latest Releases". WPExperts.io. Archived from the original on 2026-01-26. Retrieved 2026-01-26.
- ↑ "Terms of Use | OpenAI". Archived from the original on 2026-01-27. Retrieved 2026-01-27.
{{cite web}}:|archive-date=/|archive-url=timestamp mismatch; 2026-01-23 suggested (help) - ↑ "The ChatGPT home page". OpenAI Help Center. Retrieved 2026-02-19.
{{cite web}}: CS1 maint: url-status (link) - ↑ "Model Release Notes". OpenAI Help Center. Retrieved 2026-02-19.
- ↑ "ChatGPT Enterprise and Edu - Models & Limits". OpenAI Help Center. Retrieved 2026-02-19.
- ↑ "Data Usage for Consumer Services FAQ | OpenAI Help Center". Archived from the original on 2026-01-27. Retrieved 2026-01-27.
- ↑ Simo, Fidgi (20 March 2023). "March 20 ChatGPT outage: Here's what happened". OpenAI. Archived from the original on 2025-08-24. Retrieved 26 Jan 2026.
- ↑ OpenAI (13 May 2024). "Hello, GPT-4o". OpenAI. Archived from the original on 16 Dec 2025. Retrieved 2026-02-17.
- ↑ Franzen, Carl (30 Apr 2025). "OpenAI rolls back ChatGPT's sycophancy and explains what went wrong". VentureBeat. Archived from the original on 15 Aug 2025. Retrieved 15 Feb 2026.
- ↑ Hill, Kashmir (13 Jun 2025). "They Asked an A.I. Chatbot Questions. The Answers Sent Them Spiraling". The New York Times. Archived from the original on 17 Jan 2026. Retrieved 2025-02-15.
- ↑ Bradbury, Danny (6 Jun 2025). "OpenAI forced to preserve ChatGPT chats". Malwarebytes. Archived from the original on 14 Nov 2025. Retrieved 26 Jan 2026.
- ↑ R, Sudha (15 Jul 2025). "ChatGPT To Retain Conversations Indefinitely". Medium. Retrieved 26 Jan 2026.
{{cite web}}: CS1 maint: url-status (link) - ↑ Belanger, Ashley (10 Oct 2025). "OpenAI will stop saving most ChatGPT users' deleted chats". Ars Technica. Archived from the original on 1 Jan 2026. Retrieved 26 Jan 2026.
- ↑ Edwards, Benj (11 Aug 2025). "The GPT-5 rollout has been a big mess". Ars Technica. Archived from the original on 9 Jan 2026. Retrieved 5 Feb 2026.
- ↑ Franzen, Carl (8 Aug 2025). "OpenAI returns old models to ChatGPT as Sam Altman admits 'bumpy' GPT-5 rollout". VentureBeat. Archived from the original on 2026-02-05. Retrieved 5 Feb 2026.
- ↑ "Age prediction in ChatGPT". OpenAI. 21 Jan 2026. Archived from the original on 19 Jan 2026. Retrieved 26 Jan 2026.
- ↑ Simo, Fidgi (16 Jan 2026). "Our approach to advertising and expanding access to ChatGPT". OpenAI. Archived from the original on 18 Jan 2026. Retrieved 25 Jan 2026.
- ↑ Cirello, Rafaelle F; Backholer, Kathryn (23 Jan 2026). "OpenAI will put ads in ChatGPT. This opens a new door for dangerous influence". Archived from the original on 25 Jan 2026. Retrieved 26 Jan 2026.
- ↑ https://x.com/OpenAI/status/2020936703763153010
- ↑ OpenAI (9 Feb 2026). "Testing ads in ChatGPT". OpenAI. Archived from the original on 2026-02-13. Retrieved 13 Feb 2026.