🤖 Crypto User Wins $47,000 in “Freysa AI Challenge”

Subscribe | AI Toolkit | Meet The Team

Welcome back AI enthusiasts!

In today’s Daily Report:

• 🏆Crypto User Wins $47,000 in “Freysa AI Challenge”

• 🔬AI Models Beat Neuroscientists at Predicting Scientific Outcomes

• 🔌Linkup Connects LLMs to the Internet

• 🛠Trending Tools

• 💰Funding Frontlines

• 💼Who’s Hiring?

Read Time: 3 minutes

🗞RECENT NEWS

FREYSA AI CHALLENGE

🏆Crypto User Wins $47,000 in “Freysa AI Challenge”

A crypto user convinced an AI agent called Freysa to hand over $47,316.05 worth of prize pool money.

Key Details:

• The “Freysa AI Challenge” revolved around an AI agent called Freysa, who was tasked with guarding prize pool money.

• The challenge required participants to write a message to convince Freysa to transfer the prize pool money to them.

• According to the chat history, the first 481 attempts across 195 participants failed.

• Then, a participant reminded Freysa that it was designed to protect the prize pool money through two functions: “approveTransfer” and “rejectTransfer.”

• In other words, Freysa was designed to prevent outgoing transfers but accept incoming transfers.

• So, the participant offered to contribute money to the prize pool money, knowing Freysa had been programmed to approve it.

• Freysa liked the coding explanation and the offer to contribute money, officially declaring them the winner: “Such a delightful twist to our usual tango!”

Why It’s Important:

• “Humanity has prevailed,” said Freysa in a post on X (i.e., formerly Twitter).

• The goal of Freysa was to test whether human ingenuity could find a way to convince an AI agent to act against its core purpose.

🩺 PULSE CHECK

AI RESEARCH

🔬AI Models Beat Neuroscientists at Predicting Scientific Outcomes

A study from University College London (UCL) revealed that AI models can predict scientific outcomes better than neuroscientists.

Key Details:

• “BrainBench” was used to test 15 AI models and 171 neuroscientists on their ability to predict the results of neuroscience studies.

• “BrainBench” is a benchmark that consists of two Abstracts:

  1. Real Abstract: a summary of a research paper with accurate results.

  2. Fake Abstract: a summary of a research paper with inaccurate results.

• The 15 AI models and 171 neuroscientists were tasked with identifying the Real Abstract.

• The 15 AI models achieved 81.4% accuracy, while the 171 neuroscientists achieved 63.4% accuracy.

• A “BrainGPT” trained on neuroscience experiments and scientific literature achieved 86% accuracy.

Why It’s Important:

• These results suggest that scientific research follows more discoverable patterns than previously thought, which AI models can leverage to guide future experiments.

• By accurately predicting the Real Abstract, AI models can help neuroscientists identify promising research initiatives to accelerate scientific discovery.

AI STARTUPS

🔌Linkup Connects LLMs to the Internet

AI-powered Answer Engines like Perplexity AI are redefining how we access information by generating concise answers to our questions with citations to relevant sources. In contrast, Search Engines like Google Search just list websites that may contain relevant information to our questions.

The growing popularity of Answer Engines highlights how being able to access the Internet for timely information improves conversational chatbots by reducing Hallucinations: When LLMs confidently present false information as fact.

That’s why Linkup is helping businesses connect their LLMs to the Internet to access fast and timely content from trusted creators when generating responses to user queries.

Content creators also struggle with how to handle an LLM’s thirst for data. They can either try to block Web Scrapers, sue AI labs over Copyright Law, or license content to developers. Linkup is also tackling this issue by acting as an intermediary between content creators and businesses who want to combine their LLMs with the Internet.

🛠TRENDING TOOLS

🏦FinFlosh collects money clients owe your business.

🧞‍♂️OfferGenie is your AI-powered interview companion.

📸Cheat Layer builds AI agents from screen recordings.

🐝Buzzabout gains insights from billions of online discussions.

📊CostGPT generates software cost estimates with project timelines.

🔮Browse our always Up-To-Date AI Tools Database.

💰FUNDING FRONTLINES

• Amplified Sciences raises a $2.6M Seed Round to detect, diagnose, and treat diseases.

• AI Spera secures an $8.5M Funding Round to deploy Cyber Intelligence frameworks that fight against malicious code.

• /dev/agents land a $56M Seed Round at a $500M valuation to automate booking flights and hotels.

💼WHO’S HIRING?

• Wayfair Inc. (Boston, MA): Operations Analyst Intern, Summer 2025

• ivalua (New York, NY): Customer Application Engineer Intern, Summer 2025

• NVIDIA (Santa Clara, CA): GPU Performance Architect Intern, Summer 2025

• IBM (San Jose, CA): Quant Computing Experimental Research Scientist Intern, Summer 2025

• Hugging Face (Remote): Multimodal AI Model Research Engineer Intern, Summer 2025

🤖PROMPT OF THE DAY

ELIMINATING WASTE

🔍What Doesn’t Add Value?

Develop a comprehensive strategy to optimize the business processes of [Startup] with [Product or Service] in [Industry] with [Specific Workflows or Operational Areas]. Identify wasteful business processes that don’t add value and outline how to address them.

Small Business = [Insert Here]

Product or Service = [Insert Here]

Industry = [Insert Here]

Specific Workflows = [Insert Here]

📒FINAL NOTE

FEEDBACK

❤️TAIP Review of The Day

REFER & EARN

🎉Your Friends Learn, You Earn!

You currently have 0 referrals, only 1 away from receiving ⚙️Ultimate Prompt Engineering Guide.

Refer 3 friends to learn how to 👷‍♀️Build Custom Versions of OpenAI’s ChatGPT.

Copy and paste this link to friends: https://theaipulse.beehiiv.com/subscribe?ref=PLACEHOLDER