Meet Anthropic's new and improved LLM

ALSO an update on AI and the 2024 election

Welcome back to The Daily Update — Happy Friday! It’s been an eventful week in AI and we have more significant news on deck today. Here’s the full menu:

🗞️ The Latest in AI: FEC could limit AI in political ads ahead of 2024

🤖 AI Tool of the Day: Claude Instant 1.2 — Anthropic’s new and improved LLM

📸 Business Spotlight: News outlets have new demands for AI training data

🚨 AI Roundup: Four quick hits

Read time: 2 minutes

🗞️ The Latest in AI: FEC could limit AI in political ads ahead of 2024

Fake AI images of Trump embracing Fauci used by the DeSantis campaign team. (Source: The Verge)

The Federal Election Commission decided that it may establish rules regulating the use of AI-generated content in political ads. On Thursday the FEC voted to open up a petition filed by the advocacy group Public Citizen for public comment.

What you need to know:

  • New rules governing how campaigns use AI could go into effect before the end of the year.

  • The petition calls on the commission to punish fraud by creating rules banning candidates and political parties from using AI to misrepresent their opponents.

  • Thursday’s vote comes as Congress and the White House push to regulate AI.

  • Political groups like the Republican National Committee and Never Back Down, a Ron DeSantis super PAC, have already started using AI technology.

  • Once the petition hits the federal register, the public will have 60 days to comment.

Why it matters: AI is already being used to generate content that misleads voters. With debates and primaries right around the corner, government regulation might be the only way to prevent AI-generated content from interfering with the 2024 election.

Click here to learn more.

🤖 AI Tool of the Day: Claude Instant 1.2 — Anthropic’s new and improved LLM

Anthropic, the AI startup co-founded by ex-OpenAI execs, has released an updated version of Claude Instant. Claude Instant 1.2 is Anthropic’s fastest model with strength in creative tasks.

If you’ve been keeping up with this newsletter for a while, you know that I think Claude 2 (Anthropic’s flagship model) is a cut above ChatGPT. Claude Instant 1.2 incorporates the strengths of Claude 2 and improves on them.

Why you should try Claude Instant:

  • It generates longer, more structured responses and follows formatting instructions better.

  • The model shows improvement in quote extraction, multilingual capabilities and question answering.

  • Claude Instant 1.2 is less likely to hallucinate (generate text that is false or nonsensical) and more resistant to jailbreaking.

  • The model shows significant gains in math, coding, reasoning and safety.

Quora delivers access to Claude Instant through its generative AI app Poe. Click here to try it out. You’ll need a subscription for unlimited access to the most recent version of Claude Instant, featuring a context window of 100,000 tokens.

📸 Business Spotlight: News outlets have new demands for AI training data

Images generated by Midjourney

Several news organizations released an open letter calling for more transparency and copyright protection in AI.

Key points:

  • The letter urges lawmakers around the globe to consider regulations that require transparency into training datasets and consent of rights holders before using data for training.

  • Signatories want AI companies to eliminate bias and misinformation in their services. They also request that media companies identify AI-generated content.

  • Notable signatories include the Associated Press, Getty Images, the National Writers Union, the National Press Photographers Association and several European news agencies.

Why it matters: A few weeks ago Google reportedly demonstrated its generative AI news writing tool Genesis to The New York Times, The Washington Post and Wall Street Journal owner News Corp. It won’t be long before AI is writing news, but this could open a Pandora’s box of misinformation and copyright issues if done recklessly.

Click here to learn more.

🚨 AI Roundup: Four quick hits

  • Science: AI is building highly effective antibodies that humans would have never found.

  • China: Tech giants place $5 billion order for Nvidia AI chips.

  • Online retail: Amazon offers sellers AI tool to write product descriptions.

  • Personalized AI: ChatGPT expands ‘custom instructions’ feature to free users.

Have questions or thoughts on today’s newsletter? Reply to this email and I’ll get back to you as soon as I can. Thanks for reading The Daily Update!

Jack

(P.S. If you know someone that may find today’s newsletter useful, send it their way!)