Twitter Blames Microsoft For Data Misuse in Potential Future Conflict Over AI

Twitter Blames Microsoft For Data Misuse in Potential Future Conflict Over AI
The shadows of people holding mobile phones are cast onto a backdrop projected with the Twitter logo in an illustration picture taken in Warsaw, Poland, on Sept. 27, 2013. (Kacper Pempel/Reuters)
Naveen Athrappully
5/19/2023
Updated:
5/19/2023
0:00

Twitter is accusing Microsoft of misusing its API data access in a move that could foreshadow a potential conflict over the data access of AI systems.

In a May 18 letter (pdf) to Microsoft CEO Satya Nadella, Twitter said that Microsoft “may have been in violation” of multiple provisions of the Twitter Developer Agreement for an “extended period of time.” Such violations include “excessive or abusive” use of Twitter APIs, using it for “unauthorized uses and purposes” like sharing data with government agencies, and utilizing the APIs for “expressly prohibited use cases.”

“For years, Microsoft has used Twitter’s standard developer APIs free of charge in order to benefit from Twitter’s data and services in key Microsoft products that generate tens of billions of dollars in revenue for Microsoft annually.”

But in April, Microsoft stopped using these APIs after Twitter began charging for the access. Microsoft refused to pay “even a discounted rate for continued access to Twitter’s APIs and content,” the letter said.

Until then, Microsoft is said to have operated eight separate Twitter API apps that provided data and functionality to at least five Microsoft products and services—Xbox One, Ads, Bing Pages, Power Platform, and Azure.

Twitter’s letter comes as companies with large swathes of data attempt to stop AI systems from free or cheap access to such data. As big tech firms race to build AI models, data ownership is increasingly becoming a conflict point.

Artificial intelligence systems require large sets of data for training purposes. Much of this data is freely taken from online sources like Twitter and Reddit. Data from such social media sites are extremely useful for training AI due to the informal conversations they contain.

Data owners are either seeking to stop AIs from accessing their content or fairly charging companies behind the AI for such access.

In April, Musk threatened to sue Microsoft as it “trained illegally using Twitter data,” apparently referring to data used in training chatbots like Open AI’s ChatGPT and Microsoft’s Bing.

“I’m open to ideas, but ripping off the Twitter database, demonetizing it (removing ads), and then selling our data to others isn’t a winning solution,” Musk said in an April 20 tweet.
Earlier in December, Musk had blocked OpenAI’s access to Twitter data. Even though OpenAI was paying $2 million annually for the access, the billionaire believed the amount was unfair. In April, Reddit announced putting its API access behind a paywall.

Microsoft’s Violations

In its letter to Microsoft, Twitter pointed out that there were rate limits on the use of its APIs, which the agreement expressly prohibits developers from “circumventing” or “exceeding.” In addition, using APIs in a manner that “exceeds reasonable request volume” or “constitutes excessive or abusive usage” is also prohibited.

“Despite these limitations, the Microsoft Apps accessed Twitter’s APIs over 780 million times and retrieved over 26 billion tweets in 2022 alone. Indeed, for one of the Microsoft Apps, Microsoft’s account information outright states that it intends to allow its customers to ‘go around throttling limits,’” the letter said.

Twitter blames Microsoft for using APIs for “unauthorized uses and purposes.” The agreement requires that Microsoft discloses and obtains approval for how it will use the Twitter API for each Microsoft app. “Yet Microsoft did not identify any use case for six of the eight Microsoft Apps that it continued to operate until last month.”

Twitter insists that a few Microsoft Apps “appear to have engaged in expressly prohibited use cases.” For instance, Microsoft’s Azure Logic Apps supplied Twitter content to several Microsoft endpoints which referenced a government entity.

This violated the agreement as Twitter’s rules prohibit retrieval of its data for “any government-related entity” without first identifying such end users and informing about the same to Twitter. However, Microsoft did not notify Twitter about these matters, the letter claims.

“Several of the Microsoft Apps were also used in Microsoft applications that include automation capabilities, including Power Platform and Ads. Per the terms of the Agreement, Microsoft was obligated to comply with Twitter’s Automation Rules in its operation of these apps,” Twitter said.

“Yet Microsoft’s API requests for these apps include Twitter actions that are subject to certain restrictions on automation, including retweets and direct messages.”

Twitter is asking that Microsoft agree to a “compliance audit” in accordance with the Developer Agreement. The audit is to be conducted on eight Microsoft Apps through April 2023. The company gave Microsoft a June 7 deadline to respond to the letter.

The Epoch Times reached out to Microsoft for comment.