OpenAI CEO 'feels awful' about ChatGPT leaking conversations
(2023/03/23)
- Reference: 1679602012
- News link: https://www.theregister.co.uk/2023/03/23/openai_ceo_leak/
- Source link:
OpenAI CEO Sam Altman feels "awful" about ChatGPT leaking some users' chat histories on Monday, and blamed an open source library bug for the snafu.
In a couple of tweets, Altman admitted the [1]flaw , which allowed some users to see snippets of others' conversations — not the full contents, but recent titles – with the question-and-response bot.
"We had a significant issue in ChatGPT due to a bug in an open source library, for which a fix has now been released and we have just finished validating," Altman [2]said .
[3]
"A small percentage of users were able to see the titles of other users' conversation history. We feel awful about this."
[4]
[5]
Because of the buggy code, ChatGPT users won't be able to access most of their March 20 conversations, he [6]added .
OpenAI also plans to follow up with a technical postmortem about the privacy breach, according to Altman. The formerly non-profit biz did not respond to The Register 's inquiries about which open-source library contained the buggy code, and how many users were affected.
[7]
There's no word yet on when the fix will be released and when the postmortem will publish, either.
[8]Bogus ChatGPT extension steals Facebook cookies
[9]Russian criminals can't wait to hop over OpenAI's fence, use ChatGPT for evil
[10]Google reminds everyone it too can launch a ChatGPT-like chatbot … waiting list
[11]GitHub Copilot learns new tricks, adopts this year's model
While users are understandably peeved about the conversation leaks, Kaspersky's lead data scientist told The Register that ChatGPT users should read the small print — and forget any illusion of privacy.
"ChatGPT warns on login that 'conversations may be reviewed by our AI trainers,'" Vlad Tushkanov said, noting that the web demo and the API for businesses use different interfaces. "So from the very beginning the users should have had zero expectation of privacy when using the ChatGPT web demo."
Kaspersky has the following advice, he added: "Treat any interaction with a chatbot (or any other service, for that matter) as a conversation with a complete stranger. You don't know where the content will end up, so refrain from revealing any personal or sensitive information about yourself or other people."
Meanwhile, look at the new toys!
On Thursday OpenAI announced the rollout of [12]ChatGPT plugins to connect the chatbot to third-party apps, thus allowing the chatbot to do things like order food via Instacart on behalf of the users or book a flight on Expedia.
The plugins also allow ChatGPT to access real-time information, like stock prices and sports scores, or company documents stored on your device — if you trust the chatbot with those.
"You can install plugins to help with a wide variety of tasks," Altman [13]tweeted . "We are excited to see what developers create!"
[14]
No doubt, the data thieves are, too. ®
Get our [15]Tech Resources
[1] https://twitter.com/JordanLWheeler/status/1637853083865579520
[2] https://twitter.com/sama/status/1638635717462200320
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://twitter.com/sama/status/1638635718691135488
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[8] https://www.theregister.com/2023/03/23/chatgpt_fake_chrome_extension/
[9] https://www.theregister.com/2023/01/18/russia_openai_chatgpt_workarounds/
[10] https://www.theregister.com/2023/03/21/google_bard_ai/
[11] https://www.theregister.com/2023/03/22/github_copilot_learns_new_tricks/
[12] https://openai.com/blog/chatgpt-plugins
[13] https://twitter.com/sama/status/1638951865319821312
[14] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[15] https://whitepapers.theregister.com/
In a couple of tweets, Altman admitted the [1]flaw , which allowed some users to see snippets of others' conversations — not the full contents, but recent titles – with the question-and-response bot.
"We had a significant issue in ChatGPT due to a bug in an open source library, for which a fix has now been released and we have just finished validating," Altman [2]said .
[3]
"A small percentage of users were able to see the titles of other users' conversation history. We feel awful about this."
[4]
[5]
Because of the buggy code, ChatGPT users won't be able to access most of their March 20 conversations, he [6]added .
OpenAI also plans to follow up with a technical postmortem about the privacy breach, according to Altman. The formerly non-profit biz did not respond to The Register 's inquiries about which open-source library contained the buggy code, and how many users were affected.
[7]
There's no word yet on when the fix will be released and when the postmortem will publish, either.
[8]Bogus ChatGPT extension steals Facebook cookies
[9]Russian criminals can't wait to hop over OpenAI's fence, use ChatGPT for evil
[10]Google reminds everyone it too can launch a ChatGPT-like chatbot … waiting list
[11]GitHub Copilot learns new tricks, adopts this year's model
While users are understandably peeved about the conversation leaks, Kaspersky's lead data scientist told The Register that ChatGPT users should read the small print — and forget any illusion of privacy.
"ChatGPT warns on login that 'conversations may be reviewed by our AI trainers,'" Vlad Tushkanov said, noting that the web demo and the API for businesses use different interfaces. "So from the very beginning the users should have had zero expectation of privacy when using the ChatGPT web demo."
Kaspersky has the following advice, he added: "Treat any interaction with a chatbot (or any other service, for that matter) as a conversation with a complete stranger. You don't know where the content will end up, so refrain from revealing any personal or sensitive information about yourself or other people."
Meanwhile, look at the new toys!
On Thursday OpenAI announced the rollout of [12]ChatGPT plugins to connect the chatbot to third-party apps, thus allowing the chatbot to do things like order food via Instacart on behalf of the users or book a flight on Expedia.
The plugins also allow ChatGPT to access real-time information, like stock prices and sports scores, or company documents stored on your device — if you trust the chatbot with those.
"You can install plugins to help with a wide variety of tasks," Altman [13]tweeted . "We are excited to see what developers create!"
[14]
No doubt, the data thieves are, too. ®
Get our [15]Tech Resources
[1] https://twitter.com/JordanLWheeler/status/1637853083865579520
[2] https://twitter.com/sama/status/1638635717462200320
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://twitter.com/sama/status/1638635718691135488
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[8] https://www.theregister.com/2023/03/23/chatgpt_fake_chrome_extension/
[9] https://www.theregister.com/2023/01/18/russia_openai_chatgpt_workarounds/
[10] https://www.theregister.com/2023/03/21/google_bard_ai/
[11] https://www.theregister.com/2023/03/22/github_copilot_learns_new_tricks/
[12] https://openai.com/blog/chatgpt-plugins
[13] https://twitter.com/sama/status/1638951865319821312
[14] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/aiml&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33ZBzaFHQhz3fSvJ1JkTG9kAAAAIM&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[15] https://whitepapers.theregister.com/
Re: I don't think it was a bug
Filippo
> leak conversations (that is use details from previous conversations in future ones)
LLMs do lie and mislead, but they can't leak previous conversations, because they don't learn from them. They can't, their training is fixed. They appear to remember previous bits within a conversation only because the entire conversation is in their context, i.e. the whole lot gets fed to it again each time you enter a new line.
The leak discussed in the article is a plain old programming bug, not something the LLM produced as an output.
> Don't believe me ? Ask ChatGPT itself:
Believing a LLM, is, generally speaking, a poor idea.
Maybe it was a mistake...
chuckufarley
...But the profits were not.
I don't think it was a bug
A lot of these big companies diving into AI will end up bitten when it turns out they never grasped what they were doing.
"AI" in the guise of LLM will lie, leak conversations (that is use details from previous conversations in future ones), and give deliberately misleading answers. Because that is precisely what it has been trained to do.
Don't believe me ? Ask ChatGPT itself:
Are the recent examples of chatgpt disclosing the results of previous conversations a feature or bug ?
As an AI language model, ChatGPT does not have intentions or motivations and operates solely based on the input it receives and its programming.
That being said, disclosing the results of previous conversations could be seen as either a feature or a bug depending on the context and intended use case.
If the purpose of ChatGPT is to provide personalized recommendations or assistance based on previous conversations with a user, then disclosing the results of those conversations could be seen as a useful feature that improves the quality of the interaction.
On the other hand, if the previous conversations were meant to be private and not disclosed to others, then disclosing the results would be a bug and a violation of privacy.
In summary, whether disclosing previous conversations is a feature or a bug depends on the intended use case and whether the disclosure is consistent with the expectations and privacy preferences of the users.