Reuters Claims TikTok Is Working On Algorithm Copy For The U.S. - 1

Reuters Claims TikTok Is Working On Algorithm Copy For The U.S.

  • Written by Andrea Miliani Former Tech News Expert
  • Fact-Checked by Justyn Newman Former Lead Cybersecurity Editor

Anonymous sources told the news agency Reuters that TikTok is working on a duplicate of the social media’s core algorithm for its current 170 million users in the United States.

According to Reuters , employees in China and the U.S. have been ordered to separate millions of lines of code to create an independent code base that would not be connected to its Chinese version Douyin nor linked to its parent company ByteDance.

ByteDance—considered a national security threat by the U.S. government for its connection with the Chinese government— required a copy of the algorithm by the end of last year, before the US House passed a bill to ban Tiktok in the country in March. This complex task could take more than a year to be completed and has been described by the sources as “tedious dirty work”, as determining which parts of the code comply with U.S. legal requirements has made the job harder.

Reuters’ sources assure the copy project has been discussed through Lark, TikTok’s internal communication system, and that executives have shared plans and code-splitting updates. If the operation is successful, TikTok is aware that the new TikTok U.S. algorithm could not perform as well as global TikTok, optimized by ByteDance’s engineers.

The Code-Splitting Debate

After Reuters published the report, Tiktok posted a public announcement on X , saying the news agency shared false information.

“The Reuters story published today is misleading and factually inaccurate,” and added “As we said in our court filing, the ‘qualified divestiture’ demanded by the Act to allow TikTok to continue operating in the United States is simply not possible: not commercially, not technologically, not legally. And certainly not on the 270-day timeline required by the Act.”

However, Reuters trusts its sources and a spokesperson confirmed: “We stand by our reporting.” Journalist Krystal Hu, who covered the story, replied to the publication on X. “We reached out to you to seek comment before the story was published, and you responded the info was “correct”. Since publishing, I’ve been trying to reach you in every way I can but you’ve ignored me while stating our story is ‘inaccurate’. What has changed?” she wrote.

The Verge also contacted TikTok regarding the algorithm copy and spokesperson Michael Hughes replied that the information shared by Reuters is incorrect, that it’s “simply false to suggest that this work would facilitate divestiture or that divestiture is even a possibility”, and that the code copy procedure is “100 percent false.”

Anthropic’s AI Allows Users to Create Personal Assistants - 2

AI technology microchip background digital transformation concept

Anthropic’s AI Allows Users to Create Personal Assistants

  • Written by Shipra Sanganeria Cybersecurity & Tech Writer
  • Fact-Checked by Justyn Newman Former Lead Cybersecurity Editor

Anthropic announced last week that it is releasing a new feature for its AI chatbot Claude that allows users to create customized AI assistants. This feature called “tool use,” also known as “function calling,” allows Claude to connect with any external tool and API, creating a variety of personalized solutions.

According to Anthropic, the tool use feature allows Claude to automate tasks, analyze data, and create highly customized customer engagement tools. For example, Claude can be configured to create an email assistant or a bot that instantly responds to customer inquiries. It can also automate tedious tasks like data entry or file management and offer real-time technical support. However, setting up these personalized bots requires API access and coding knowledge.

Additionally, the tool’s functionality includes working with images, allowing users to incorporate image inputs in applications that analyze visual data.

The New Claude 3 Family & Notable Use Cases

The Claude 3 family includes different models, namely Haiku, Sonnet, and Opus. The Claude tool, which has been in beta phase since April, saw most of the customers opt for Anthropic’s fastest and most affordable Claude 3 Haiku model.

Furthermore, one of Anthropic’s project lead, Dianne Penn, shared with The Verge that the beta testing results suggest promising startup solutions. One of the several clients using the Claude tool during the beta was StudyFetch, which used the tool to create a personalized AI tutor called Spark.E. Some other notable mentions were Intuned, the browser automation platform and Hebbia, a startup that develops AI-infused search tools.