Rob Waugh
·Contributor
Updated Fri, 26 July 2024
Elon Musk's X network is to start training AI on user data (Photo by Marc Piasecki/Getty Images)
Elon Musk’s X, formerly known as Twitter, has added a setting which allows the network to use users’ information to train its Grok AI.
The new setting appears to be "on" by default for most users and allows the network to gather their information to train Grok - not just from posts, but from other interactions with the site.
The new setting says: "To continuously improve your experience, we may utilize your X posts as well as your user interactions, inputs and results with Grok for training and fine-tuning purposes.
"This also means that your interactions, inputs, and results may also be shared with our service provider xAI for these purposes."
The company made the change with little fanfare so it will have doubtless gone unnoticed by many users, although Musk did announce that his the company had begun training Grok this week in a post on X, promising "the world’s most powerful AI by every metric by December this year."
What is Grok?
Grok is a chatbot similar to ChatGPT only accessible to X subscribers (Getty Images)
Grok is a chatbot (similar to ChatGPT and Anthropic’s Claude) that is accessible via X, but only to subscribers.
Once known as TruthGPT, Musk initially billed Grok as "a maximum truth-seeking AI that tries to understand the nature of the universe". Musk has promised that Grok will be ‘anti-woke’ and offers a ‘Fun Mode’ as well as an ‘Unhinged Fun Mode’.
Why is the new setting controversial?
Similar changes at Meta (owner of Facebook and Instagram) have triggered complaints from data rights groups. Meta emailed Facebook and Instagram users in May to warn of a change, aiming to use user data to develop AI.
Privacy campaigners Open Rights Group (ORG) said: "With the latest update of their privacy policy, Meta has stated their intention to feed all the data they hold (about you) on their Facebook and Instagram platforms for the purposes of training 'AI technology'.
"While Meta also stated that they would only use 'public posts', their privacy policy contradicts this by including any personal data collected through these platforms, with the sole exclusion of private chats."
ORG has raised a complaint with regulators over the issue.
Veronika Pozdniakova, a privacy expert at privacy-focused phone company Murena, said, "As is often the case with online services, the new terms and conditions lack transparency, and this change directly impacts user privacy because the AI will be trained using users' personal information, which they may not be aware of or fully comprehend.
Users should regularly review and manage data shared on such platforms, including posts, interactions, and direct messages. Users also can opt out by abandoning a platform whose terms and conditions they don't agree with."
How can you stop your data being used like this?
You can’t change this setting in the mobile app, but you can withdraw consent to your data being used in the desktop version of X on Mac or PC.
To do so, just go to your profile, then Settings > Privacy and Safety > Data Sharing and Personalization.
Select Grok, and then switch the slider to ‘Off.’
Elon Musk Is Now Using Your Tweets to Train His Weird AI. Here's How to Turn It Off.
Noor Al-Sibai
Fri, 26 July 2024
Sharing Is Caring
At some point, X-formerly-Twitter auto-opted its users into sharing their data with Grok, owner Elon Musk's unfunny "anti-woke" chatbot — without bothering to tell users about the change.
Flagged by Rolling Stone's Noah Shachtman, it remains unclear when X introduced this so-called "data sharing" change that authorizes the social network to use your tweets to train Musk's shoddy ChatGPT competitor.
As Shachtman notes, opting out of having your data included in Grok's training pool is as simple as de-selecting the allowance in your settings.
Nevertheless, it's super sketchy that everyone was automatically opted in without being informed, and that nobody knows how long this has been going on.
Forgiveness, Not Permission
Beyond not informing users that it's using their data, X also appears to not have told regulators about the change, either.
In a statement to TechCrunch, a spokesperson for the Irish Data Protection Commission (DPC), which acts as X's European watchdog, said the agency was taken by surprise when it learned that the social network had automatically opted users into allowing their tweets to be used as AI training data.
"The DPC has been engaging with X on this matter for a number of months, with our latest interaction occurring as recently as yesterday," Graham Doyle, the DPC's deputy commissioner, told the site. "Therefore we are surprised by today’s developments."
As TechCrunch notes, European Union law requires companies to have a valid legal reason to process data — and per the brief explanation provided under the "data sharing" tab in X user settings, such reasoning is unclear.
The watchdog said it had followed up with X about the change and that it's waiting for "further engagement" on the matter.
With this brewing debacle, X has become the latest in a string of companies to scrape data without asking. It's also not the first to automatically opt users into data sharing as a means of ass-covering — and unfortunately, it probably won't be the last.
More on AI training: When AI Is Trained With AI-Generated Data, It Starts Spouting Gibberish
Here's how to disable X (Twitter) from using your data to train its Grok AI
Aisha Malik
Fri, 26 July 2024
X, formerly known as Twitter, has automatically activated a setting that allows the company to train its Grok AI on users’ posts. X enabled the new setting by default. The good news is that you can switch it off and also delete your conversation history with the AI.
If the setting is turned on, X can “utilize your X posts as well as your user interactions, inputs and results with Grok for training and fine-tuning purposes,” according to the platform’s settings page. X goes on to note that “this also means that your interactions, inputs, and results may also be shared with our service provider xAI for these purposes.”
Although it’s not possible to disable the setting via X’s mobile app, you can do so on the desktop version of the social network.
How to switch off X's data sharing settings:
Open up the Settings page on X on your desktop.
Select the “Privacy and safety” button.
Select “Grok.”
Uncheck the box.
A screenshot showing how to switch off the Data Sharing settings in X's desktop mode. Image Credits: TechCrunch
After you have switched off the setting, you can delete your conversation history (if any) with the AI by clicking on the “Delete conversation history” button.
Earlier this week, X owner Elon Musk said xAI had started training its Grok large language model using "the most powerful AI training cluster in the world." Musk said that the AI model would become "the world's most powerful AI by every metric by December [2024]." Now, it’s clear that Musk and X were hoping to use more than the powerful training cluster to train its AI by also using users' past tweets and posts.
X isn't the only social network that has utilized user data to train its AI, as Meta notified EU and U.K. users last month of an upcoming change that would allow it to use public content on Facebook and Instagram to train its AI. The company eventually bowed to regulatory pressure and paused its plans.
Beyond not informing users that it's using their data, X also appears to not have told regulators about the change, either.
In a statement to TechCrunch, a spokesperson for the Irish Data Protection Commission (DPC), which acts as X's European watchdog, said the agency was taken by surprise when it learned that the social network had automatically opted users into allowing their tweets to be used as AI training data.
"The DPC has been engaging with X on this matter for a number of months, with our latest interaction occurring as recently as yesterday," Graham Doyle, the DPC's deputy commissioner, told the site. "Therefore we are surprised by today’s developments."
As TechCrunch notes, European Union law requires companies to have a valid legal reason to process data — and per the brief explanation provided under the "data sharing" tab in X user settings, such reasoning is unclear.
The watchdog said it had followed up with X about the change and that it's waiting for "further engagement" on the matter.
With this brewing debacle, X has become the latest in a string of companies to scrape data without asking. It's also not the first to automatically opt users into data sharing as a means of ass-covering — and unfortunately, it probably won't be the last.
More on AI training: When AI Is Trained With AI-Generated Data, It Starts Spouting Gibberish
Here's how to disable X (Twitter) from using your data to train its Grok AI
Aisha Malik
Fri, 26 July 2024
X, formerly known as Twitter, has automatically activated a setting that allows the company to train its Grok AI on users’ posts. X enabled the new setting by default. The good news is that you can switch it off and also delete your conversation history with the AI.
If the setting is turned on, X can “utilize your X posts as well as your user interactions, inputs and results with Grok for training and fine-tuning purposes,” according to the platform’s settings page. X goes on to note that “this also means that your interactions, inputs, and results may also be shared with our service provider xAI for these purposes.”
Although it’s not possible to disable the setting via X’s mobile app, you can do so on the desktop version of the social network.
How to switch off X's data sharing settings:
Open up the Settings page on X on your desktop.
Select the “Privacy and safety” button.
Select “Grok.”
Uncheck the box.
A screenshot showing how to switch off the Data Sharing settings in X's desktop mode. Image Credits: TechCrunch
After you have switched off the setting, you can delete your conversation history (if any) with the AI by clicking on the “Delete conversation history” button.
Earlier this week, X owner Elon Musk said xAI had started training its Grok large language model using "the most powerful AI training cluster in the world." Musk said that the AI model would become "the world's most powerful AI by every metric by December [2024]." Now, it’s clear that Musk and X were hoping to use more than the powerful training cluster to train its AI by also using users' past tweets and posts.
X isn't the only social network that has utilized user data to train its AI, as Meta notified EU and U.K. users last month of an upcoming change that would allow it to use public content on Facebook and Instagram to train its AI. The company eventually bowed to regulatory pressure and paused its plans.
No comments:
Post a Comment