Get all your news in one place.
100’s of premium titles.
One app.
Start reading
The Street
The Street
Ian Krietzberg

Tech News Now: A tough road ahead for TikTok, Inflection's new model, and more

Fast Facts

  • House Energy and Commerce Committee voted 50-0 Thursday to approve a bill to force ByteDance to divest TikTok or face a ban. 
  • AI startup Inflection launched Inflection-2.5, an upgraded version of its model that competes with GPT-4. 
  • A report from Patronus AI says ChatGPT produces copyright-infringing output more than any other model. 

Good morning, happy Friday, and welcome to Tech News Now, TheStreet's daily tech rundown. 

In today's edition, we're covering the U.S. government's battle against TikTok, Inflection's newest release, copyright-infringing material generated by artificial-intelligence chatbots and a new price target from Wedbush's Dan Ives for his "Messi of AI," Palantir.

Tickers we're watching today:  (PLTR) (AAPL) (GOOG)

Don't Miss: After claims of self-awareness in Anthropic's new Claude model surfaced on X, I broke down the consciousness problem that has been plaguing philosophers for centuries, and cognitive and computer scientists for decades. Likely the real culprit of Claude's alleged self-awareness was a pattern in its training data, though transparency on that front remains lacking. 

Eight more hours until the weekend. Let's get into it. 

Related: No, Elon Musk, AI self-awareness is not 'inevitable'

The U.S. government doesn't like TikTok

Attempts to ban the popular social-media app TikTok, which boasts around 170 million U.S. users, have been ongoing for years

The government is taking another crack at it. 

In a 50-0 vote Thursday, the House Energy and Commerce Committee approved a bill that would give TikTok's Chinese parent, ByteDance, 180 days to divest TikTok or the app will be banned. 

House Majority Leader Steve Scalise (R-Louisiana) said on X that he would be bringing the bill to the full House for a vote next week. 

"This legislation has a predetermined outcome: a total ban of TikTok in the United States," TikTok said after the vote. "The government is attempting to strip 170 million Americans of their Constitutional right to free expression. 

"This will damage millions of businesses, deny artists an audience, and destroy the livelihoods of countless creators across the country."

And after warning on the app that Congress was “planning a total ban of TikTok," users flooded lawmakers with calls urging them not to support the bill. 

“This morning, prior to our hearing, TikTok used its influence and power to force users to contact their representatives if they wanted to continue using TikTok," Chairwoman Cathy McMorris Rodgers said. "This is just a small taste of how the CCP weaponizes applications it controls to manipulate tens of millions of people to further its agenda."

TikTok has said often that it does not share and has not shared user data with the Chinese government. The app doesn't seem to collect any more data than its peers. And opponents of efforts to ban the app in the past have said that user data from U.S-owned companies could easily be sold to the Chinese government through data brokers. 

Fight for the Future, a nonprofit digital-rights group, has urged lawmakers to "actually protect my sensitive data from China and other governments. Stop feeding moral panic and pass a real data-privacy law to stop Big Tech companies — including TikTok! — from harvesting and abusing our personal data for profit.”

Related: Montana declares war on one popular social media app

Inflection's fresh-baked Pi

Inflection.AI on Thursday released Inflection -2.5, an upgraded version of its AI model that the company said is "competitive with all the world's leading Large Language Models like GPT-4 and Gemini."

The upgraded version became available to all users of Pi, Inflection's personal AI chatbot, on Thursday. Pi, unlike GPT-4, Claude Opus, Gemini and Copilot, is free. 

Inflection said that it achieved parity with GPT-4 with more efficiency than OpenAI, using "only 40% of the amount of compute for training."

Inflection said Pi has a million daily users and six million monthly active users, which have exchanged a total of four billion messages with Pi. 

Like OpenAI, Inflection is backed by Microsoft  (MSFT) ; the company unveiled a $1.3 billion funding round last year that was led by the tech giant. 

I messed around with Pi for about an hour yesterday. And while Pi and I had an interesting conversation about music, there's something strange about a chatbot designed to answer questions that includes the caveat that it "may make mistakes, please don't rely on its information."

A screenshot from tech reporter Ian Krietzberg's conversation with Pi. 

Inflection 2.5 Pi

When I asked it to summarize an article I had written, it completely invented quotes from the article, then attributed them to me. 

Related: Microsoft engineer says company asked him to delete an alarming discovery

Report: ChatGPT is the biggest copyright culprit

After releasing its Copyright Catcher platform, designed to detect when language models produce "exact reproductions" of content, Patronus AI published a report that found that OpenAI's GPT-4 produced copyrighted content in its output on 44% of prompts. 

For Mistral’s Mixtral-8x7B-Instruct-v0.1, that number fell to 22%.

Anthropic's Claude 2.1 and Meta's Llama-2 produced copyrighted content on 8% and 10% of prompts respectively. 

The report comes as OpenAI and its peers remain embroiled in a storm of copyright-related lawsuits, most of which have alleged rampant copyright infringement in both the inputs and outputs of AI models. 

Though OpenAI has never denied the use of copyrighted content in the training of its models, even arguing that the use of such content is necessary to build its models, OpenAI believes it is fair to use that content to train its models without compensating, crediting or asking the original creators. 

Related: Here are all the copyright lawsuits against ChatGPT-maker OpenAI

A new price target for Palantir

Wedbush's Dan Ives said in a note Friday that he was boosting his price target for Palantir to $35 from $30, based on a recent check on the company in the field. He carries an outperform rating on the data-integration giant.

"We believe the [artificial intelligence platform] foundation is becoming viewed by many US enterprises as the 'launching pad of AI use cases,'" Ives said. 

"With the AI Revolution now quickly heading towards the key use case and deployment stage, Palantir with its flagship AIP platform and myriad of customer bootcamps is in the sweet spot to monetize a tidal wave of enterprise spend now quickly hitting the shores of the tech sector."

Ives expects Palantir to "garner a meaningful share" of what he expects to be a $1 trillion global total addressable artificial-intelligence market over the next year. 

He added that the company remains "an undiscovered gem."

Related: Top Analyst Dan Ives Has New Price Target On 'Messi of AI' Palantir

The AI Corner: Biden's State of the Union

During his State of the Union address Thursday night, President Joe Biden — very succinctly — called to "ban AI voice impersonation and more," without elaborating. 

Read his speech here

His call to ban AI voice impersonations comes as incidents of AI-powered voice cloning are on the rise, supercharging cybercriminals and scam efforts

These efforts recently targeted Biden himself: A voice-cloned robocall of Biden made the rounds in January, urging voters not to participate in the New Hampshire primary. That was just the beginning of what many cybersecurity experts expect to be a very difficult election season

Contact Ian with tips and AI stories via email, ian.krietzberg@thearenagroup.net, or Signal 732-804-1223.

Related: Building trust in AI: Watermarking is only one piece of the puzzle

Sign up to read this article
Read news from 100’s of titles, curated specifically for you.
Already a member? Sign in here
Related Stories
Top stories on inkl right now
One subscription that gives you access to news from hundreds of sites
Already a member? Sign in here
Our Picks
Fourteen days free
Download the app
One app. One membership.
100+ trusted global sources.