Ai Artificial Intelligence Archive

Stability AI announces paid membership for commercial use of its models

Emilia DavidDec 19, 2023

Anthropic will help users if they get sued for copyright infringement.

The company updated its commercial terms of service for those who use Claude, its AI chatbot, saying it will not only defend customers from copyright infringement claims but also pay for settlements. The new terms go into effect on January 1st, 2024 and follow similar commitments from Microsoft, Google, and other companies.

These copyright protections won’t apply if a customer “knows or reasonably should know” they’re infringing copyright. This is consistent with what Anthropic previously said in a comment to the US Copyright Office, where it said it didn’t believe “users should be able to create outputs using Claude that infringe copyrighted works.”

Expanded legal protections and improvements to our API

[Anthropic]

Mia SatoDec 19, 2023

The AI porn bots are hallucinating.

Sex workers are experimenting with bots that mimic their look and sound, available to chat with fans 24/7. But the robot copycats are still... working out some kinks, let’s say:

[The chatbot] said, falsely, that Dee had moved several years ago to Costa Rica and lives with two “incredibly sweet and supportive” roommates. When a reporter reminded it that Dee just had a wedding, SophieAI responded: “My husband and I aren’t married per se, but we live together happ,” ending the sentence mid-word.

Creators, porn stars turn to AI doppelgangers to keep fans entertained

[The Washington Post]

Jess WeatherbedDec 19, 2023

TomTom and Microsoft are making their own K.I.T.T.

At least, that’s what it sounds like in TomTom’s description of its upcoming Azure-based Generative AI voice assistant for vehicles.

The company says it’s built on TomTom’s Digital Cockpit, and allows drivers to “converse naturally with their vehicle,” verbally control navigation and locate stops en route, and even control onboard systems like opening windows or adjusting temperatures. It’ll be demonstrated at CES in January.

A press image of a car infotainment system featuring TomTom and Microsoft’s upcoming AI assistant.

This is the only image supplied so far — it’s a far cry from Knight Rider but certainly looks more practical.

Image: Microsoft / TomTom

Pakistan’s former prime minister is using an AI voice clone to campaign from prison

Amrita KhalidDec 18, 2023

New York City council member-elect used AI to answer questions

Emma RothDec 18, 2023

Elizabeth LopattoDec 18, 2023

John Herrman is such a good blogger it’s disgusting.

“Partnerships like this should complicate this story of inevitability for companies like OpenAI, the world-beating firm that, before it takes over the economy, and before it must be stopped from taking over the entire world, must first, for some reason, pay for a big subscription to Politico Pro.”

What Do AI Companies Want With the Media?

[Intelligencer]

Emilia DavidDec 18, 2023

OpenAI split its trust and safety team, creating three separate groups taking on AI risk.

The Information reported that OpenAI has abandoned finding a replacement for trust and safety head Dave Willner, who stepped down in July. Instead, it’s replacing the division with teams dubbed Safety Systems, Superalignment, and Preparedness teams.

The company said in a blog post that Safety Systems will focus on the safe deployment of advanced AI models and artificial general intelligence while Superalignment works to align human and AI intelligence that surpasses humans, and Preparedness will do safety assessments for foundation models.

Safety Systems

[OpenAI]

Expedia wants to use AI to cut Google out of its trip-planning business

Emilia DavidDec 18, 2023

The ultimate app for reading the internet

David PierceDec 17, 2023

Wes DavisDec 16, 2023

Microsoft says its newest compact “small language model,” Phi-2, is bigger and better.

The company has been working on training AI models on much smaller data sets comprised only of “textbook-quality” data, as part of its Phi models.

Microsoft says in a research blog that Phi-2, which is about twice as big as its predecessor, Phi 1.5, continues to perform on par or better than certain larger open-source Llama 2 models, including one with 13 billion parameters.

A chat session where Phi-2 identifies why a physics calculation was incorrect.

Phi-2 identifies an error in a physics calculation.

Image: Microsoft

Alex HeathDec 15, 2023

OpenAI suspends ByteDance’s account after it used GPT to train its own AI model.

In today’s issue of Command Line, I reported that ByteDance has been violating the developer license of both Microsoft and OpenAI by using GPT-generated data to train its own, competing model in China.

After my report was published, OpenAI spokesperson Niko Felix sent the following statement confirming that ByteDance’s account has been suspended:

All API customers must adhere to our usage policies to ensure that our technology is used for good. While ByteDance’s use of our API was minimal, we have suspended their account while we further investigate. If we discover that their usage doesn’t follow these policies, we will ask them to make necessary changes or terminate their account.

As I reported, most of ByteDance’s GPT usage has been done through Microsoft’s Azure platform, not through OpenAI directly. I’ve asked Microsoft if it will follow OpenAI and suspend ByteDance’s access as well.