• JulesPerToken
  • Posts
  • Even law enforcement agencies are playing Overwatch

Even law enforcement agencies are playing Overwatch

Apr 18, 2025

🪞 Motorola taps Perplexity AI

Motorola is making a bold move by integrating Perplexity AI into its next-gen Razr smartphone, launching April 24. Perplexity’s assistant is designed to provide real-time, conversational search and summarization — basically, a smart AI that doesn’t just search but answers.

Fun Fact: This isn’t a one-off collab. Perplexity is reportedly also working with Deutsche Telekom (T-Mobile's parent company) on an “AI Phone” that might eliminate the need for most apps entirely.

🔗 The Verge

🤖 o4-mini is here - for everyone

OpenAI just dropped o4-mini, its latest lightweight multimodal model, and it’s available to everyone, even free-tier ChatGPT users. This model accepts both text and images, handles reasoning with a process called “chain-of-thought” (where it explains its steps like a math teacher), and generally brings better accuracy to basic prompts.

Fun fact: Chain-of-thought prompting was only introduced as a novel trick in 2022. Two years later, it’s baked into every major model like it's table stakes.

🔗 Wiki

Turns out “show your work” in math class makes a lot of sense

🔐 Virtue AI raises $30M

Virtue AI is focused on making sure your AI doesn’t leak your secrets. Virtue wants to help businesses use generative AI confidently by plugging security holes before they’re exploited. That means red-teaming, stress-testing, and making sure your AI doesn’t turn into a compliance nightmare.

Fun Fact: A 2023 Stanford study found over one-third of generative models could leak sensitive info they were trained on — including emails, passwords, and internal company data.

🔐 Even law enforcement agencies are playing Overwatch

Law enforcement agencies in the U.S. are using a tool called Overwatch, an AI system that creates synthetic digital personas — like fake protesters or trafficking victims — to infiltrate private chats on encrypted platforms like Signal and Discord. Developed by a company called Massive Blue, this system is being deployed near the southern border to collect intel on human trafficking and cartel activity.

What’s raising eyebrows: there have been no arrests tied to the program yet, and critics worry the system creates a dangerous precedent for digital surveillance and entrapment — especially when driven by algorithms trained to simulate “empathetic” conversations.

Fun Fact: The AI behind Overwatch was designed to generate context-aware emotional responses — meaning it can mimic trauma, stress, or vulnerability to build trust in chats.

🔗 WIRED

🇨🇳 China Embeds AI into Schools

It’s never too early to start learning AI. Which is why China’s integrating AI into curriculum from kindergarten to university. China is investing billions annually to make AI education infrastructure as ubiquitous as math and science — something Western education systems are still largely debating in think-pieces.

Fun Fact: China has launched more than 300 AI-focused educational pilot zones across the country — some of which include AI tutors that adapt content in real-time to student progress.

🔗 Reuters

Reply

or to participate.