😺 AI voice clones get politica

PLUS: Today is Grok 4 launch day!

Welcome, humans.

Our friends at DZone are putting together their 2025 Data Engineering Trend Report, and they want to hear from the people actually building this stuff (not just the executives who talk about it in LinkedIn posts).

They're asking about the real deal—data pipelines, AI workflows, DataOps, and whether all this “data-driven culture” talk is actually happening or just fancy PowerPoint slides.

The survey takes about 12 minutes, but they’re offering two lucky participants will snag $120 gift cards for their trouble (that could be you!!).

Honestly, it's refreshing to see someone ask developers what's actually working instead of just regurgitating vendor marketing. If you're knee-deep in data engineering, your input could help shape what the industry talks about next year.

Take the survey here

Here’s what you need to know about AI today:

AI voice clones are the hot new spy tool.
OpenAI launched a $10M program to train 400K teachers in AI.
AI companies captured nearly half of Q2 2025's $91B venture funding.
Mistral could raise $1B for its open source AI.

P.S: If you use Gmail and this email gets cut off, click: “[Message clipped] View entire message” so you don’t miss anything!

AI Voice Clones Just Called Congress. Marco Rubio Was the Bait.

While no images of the perpetrator were available, we assume he looked like this…

Okay, this is a big deal. Someone just used an AI voice clone to impersonate the U.S. Secretary of State. And it worked.

According to the WashingtonPost, in mid-June 2025, attackers successfully contacted five high-level officials using an AI-generated voice clone of Marco Rubio over Signal to try accessing sensitive information. They were:

A U.S. governor.
A member of Congress.
And THREE foreign ministers.

…and the perpetrators needed only 15-20 seconds of publicly available audio to create the fake.

We’ve known for awhile now AI voice clones are a big deal. The FBI has warned of an active campaign targeting senior U.S. officials since April, using AI voices to gain access to sensitive accounts. AI voice clone heists have been a thing for awhile, too:

A UK energy company lost $243K in 2019 when the CEO received a fake call from his “boss” requesting an urgent transfer.
Criminals stole $35M from a UAE bank in 2020 using a cloned director's voice.
Most recently, a Hong Kong company lost $25.6M after a finance worker attended a video call where everyone except him was a deepfake.

Deloitte projected that AI deepfake fraud could lead to $40B in losses by 2027.

Here's what keeps security experts up at night: Voice cloning now costs as little as $1-5 per month and requires only 3 seconds of audio. Testing shows 80% of AI tools successfully clone political voices despite supposed safeguards.

And yet, humans can only distinguish real from cloned voices 50% of the time (basically a coin flip).

Here’s what’s being done: U.S President Trump signed the Take It Down Act criminalizing AI deepfakes, while Denmark proposed giving citizens copyright over their own voice for even more protection. But the most important question is this: do you have a catch phrase and/or signal to use with your loved ones to confirm it’s them?

If you don’t, you should. The question isn't whether AI voice cloning will be used against you—it's when, and whether you'll be ready

FROM OUR PARTNERS

AI That Knows Your Work Inside and Out

Most AI tools start from scratch every time. ClickUp Brain already knows the answers.

It has full context of all your work—docs, tasks, chats, files, and more. No uploading. No explaining. No repetitive prompting.

ClickUp Brain creates tasks for your projects, writes updates in your voice, and answers questions with your team's institutional knowledge built in.

It's not just another AI tool. It's the first AI that actually understands your workflow because it lives where your work happens.

Join 150,000+ teams and save 1 day per week.

Try ClickUp Brain Free

Prompt Tip of the Day

Balaji shared a prompt tip that could save you thousands on consulting fees. He suggests you can turn any professional into an AI fact-checker.

Here's the hack: First, prompt your AI to become the best lawyer/accountant/doctor in your city, then write a detailed memo about your situation with specific questions (don't forget to ask “what am I overlooking?”).

Run this same prompt across multiple AIs: ChatGPT, Claude, Gemini. Then, organize their answers in a spreadsheet.

Now, you've got a comprehensive survey of expert opinions, complete with citations and draft documents. Now take this research doc to the actual human professional and ask them to verify the AI's work.

The result? You're no longer paying billable hours for basic education. Instead, you're paying for expert verification and final polish; exactly what professionals should be doing.

Our favorite insight: This works with any expert, from doctors to accountants. AI handles the “middle-to-middle” work, while humans do what they do best: verify accuracy and add final expertise.

Check out all of our Prompt Tips of the Day from June here.

P.S: Completely new to AI? Start here!

Treats To Try.

*Asterisk = from our partners. Advertise in The Neuron here.

*Luma AI turns your text into videos and lets you completely restyle any video's background, characters, or setting in post-production.
Genspark SuperAgent researches topics into custom pages and makes actual phone calls for you—like having an assistant handle your dentist appointment while you work (Genspark is getting so popular, OpenAI wrote about them).
OlMo 2 gives you a ChatGPT-level language model that's completely open-source, including all training data and code (though some say Mistral Small 3.2 24B offers better performance with fewer parameters).
SmolLM3 is a small language model that can switch between quick responses and detailed reasoning modes, handling up to 128k context in 6 languages while outperforming other 3B models.
Clueso converts your raw screen recordings into product videos and step-by-step docs by automatically adding professional voiceovers, zoom effects, and branding to create polished videos—free trial, then paid plans.
EarthGPT interprets satellite imagery through natural language conversations, allowing anyone to analyze remote sensing data, detect objects, and understand complex Earth scenes without technical expertise.
Menu, please! helps you order Cantonese, Chinese, Japanese, Korean, Singaporean, Taiwanese, and Thai food by photographing menus to understand each dish's ingredients and flavors, then building your order in the app to show or play as translated audio to restaurant staff.

See our top 51 AI Tools for Business here!

Around the Horn.

Sam Altman just gave a man on the street interview (NOT generated by Veo 3) outside the Sun Valley “billionaire summer camp” conference and TBH, we don’t know what’s funnier: this response to a question about his relationship with Elon, or the response in the comments to those, ahem, “unique” glasses…

xAI, Elon’s AI company, will release Grok 4 today (July 9th) via livestream at 8PM PT (here’s the leaked benchmarks); it’s largely expected for OpenAI to release GPT-5 sometime after Grok 4, so say Thursday July 10th at the earliest.
OpenAI strikes back—nabbing Meta’s Angela Fan alongside Tesla and xAI heavies for its scaling team—just after Zuckerberg swiped seven researchers from OpenAI, proving the AI talent tug-of-war never ends
Mistral, the French open-source AI leader behind Le Chat, could be about to raise $1B from Abu Dhabi’s MGX fund, and LangChain (AI agent tool) might soon be worth $1B, as its pivot to “LangSmith” helped it secure enterprise customers like Klarna and Replit. Speaking of…
Replit partnered with Microsoft to offer its vibe coding platform through Azure Marketplace, which was a big blow to previous partner Google Cloud.
OpenAI partnered with the American Federation of Teachers to launch the National Academy for AI Instruction, a five-year, $10M initiative to train 400K K-12 educators (1 in 10 US teachers) to lead AI integration in classrooms rather than having AI replace human teaching.
Adobe says this year’s Prime Day could feature up to 3,200% more traffic from genAI sources as “budget-conscious consumers” leverage it to find deals.
Crunchbase reported global venture funding reached $91B in Q2 2025, with AI companies capturing nearly half of all capital.
OpenAI cracked down on security at the company by limiting access to sensitive AI models and new products, hosting them on offline systems, and even requiring fingerprint scans for certain areas (plus a beefed up physical presence at its data centers).

For the latest AI deep dives, check out our Explainer articles here!

Midweek Wisdom

AI is driving massive data industry consolidation, where companies are abandoning fragmented tools for unified platforms, as shown by Databricks' $1B Neon acquisition and Salesforce's $8B Informatica deal.
The creator of Soundslice built an ASCII tab import feature simply because ChatGPT was incorrectly telling users it existed, revealing how AI hallucinations might increasingly shape product roadmaps as users trust AI-generated misinformation.
Ethan Mollick published a new essay “against brain damage” (and misinterpretations of the MIT study), defending the use of AI in education.
Fireship explained the Soham Parekh “overemployed” debacle (the developer who worked for at least 4 startups in San Francisco at once) and how its a symptom of the larger job market’s sickness.
Sakana AI revealed a new breakthrough of letting different language models work as a "dream team" through its TreeQuest framework, improving performance by 30% while demonstrating how AI systems achieve collective superintelligence.
This AgentCompany benchmark reveals current AI agents can autonomously handle up to 30% of simulated workplace tasks—suggesting we should focus on task-level automation rather than assuming entire professions will be automated (for now).
The Algorithmic Bridge gave a great AI industry critique warning about growing dependency on fundamentally flawed AI tools, while questioning whether the industry prioritizes solving core problems like hallucinations or simply races ahead.