China is trying to control AI

The U.S. Treasury once printed $100,000 bills with Woodrow Wilson's face, used only for transactions between Federal Reserve Banks.

Love or hate this intro? Hit reply and tell us why!

What’s in store:

Read Time: 4 minutes

The Chinese Communist Party has ramped up regulations to ensure tech companies align with its ideological principles.

All AI firms must undergo government reviews to confirm their LLMs reflect “core socialist values”, as reported by the Financial Times.

China, known for its “Great Firewall” that blocks content deemed harmful to the CCP, is now extending its censorship to AI technologies.

The Cyberspace Administration of China mandates that AI companies like ByteDance and Moonshot participate in a review process to censor sensitive keywords and topics, particularly human rights-related ones.

Here’s what you should know:

These AI systems respond to restricted queries with phrases like “try different questions” or “I have not yet learned how to answer their question”.

To avoid excessive blocking, policies state that LLMs should not reject more than 5% of all questions.

Instead, politically correct blanket answers are used to handle sensitive topics.

Arthur Herman, a senior fellow at Hudson Institute, warns that China’s control over information is a significant threat.

He points out the potential for AI applications like TikTok to manipulate global populations, indicating a broader strategy of brainwashing through technology.

Where’s the prompt hackers when you need them?

Vote for live results and see results + opinions from yesterday at the bottom of the email.

This incredible 3-hour Crash Course on AI & ChatGPT (worth $399) designed for founders & entrepreneurs will help you 10x your business, revenue, team management & more.

It has been taken by 1 Million+ founders & entrepreneurs across the globe, who have been able to:

July 29th 1973 - Led Zeppelin has $203,000 in cash stolen from a safe deposit box at the New York Hilton, which was their earnings from concerts at Madison Square Garden.

BlueDot*: AI note-taker that records your meetings without bots joining calls. Integrates seamlessly into your meetings. Install Chrome Extension →

AI Username Generator: This tool provides a fun way to generate unique usernames by rearranging the letters of your input to create memorable and secure usernames.

Murf.ai: An AI voice generator that converts text to speech in up to 20 languages, featuring a range of voices and styles suitable for various content needs.

Aiva: Known as the Artificial Intelligence Virtual Artist, Aiva composes emotional soundtracks for films, video games, and advertisements.

Kalendar.AI: An innovative AI platform designed for email outreach and B2B lead generation.

The National Institute of Standards and Technology (NIST), a U.S. Commerce Department agency, has updated Dioptra, a tool to measure the impact of malicious attacks, especially those that “poison” AI training, on AI performance.

Dioptra, first launched in 2022, is a modular, open-source, web-based tool aimed at helping companies and users assess, analyse, and track AI risks.

It allows for benchmarking and research, providing a common platform for exposing models to simulated threats in a “red-teaming” environment.

According to NIST, Dioptra supports government agencies and businesses in evaluating AI developers’ claims about their systems’ performance.

Here’s what they’re doing:

Diopotra’s development aligns with the U.S.-U.K. partnership for AI model testing, established during the U.K.’s AI Safety Summit at Bletchley Park.

The tool accompanies documents from NIST and the AI Safety Institute, suggesting ways to mitigate AI dangers, such as misuse in generating non-consensual content.

Dioptra’s creation is also part of President Joe Biden’s executive order on AI, setting standards for AI safety and security.

This order requires companies like Apple to notify the federal government and share safety test results before public deployment.

While AI benchmarks are challenging due to the secretive nature of sophisticated AI models, NIST proposes that Dioptra can identify attacks that degrade AI performance and quantify their impact.

However, Dioptra currently works only on models that can be downloaded and used locally, like Meta’s Llama family, not those behind API’s like OpenAI’s GPT-4.

Guardians of the AI. (We know it’s cheesy but we’re keeping it)

Dogecoin (DOGE) saw a 7% price increase last week, briefly dipped to %0.12, and then rebounded to $0.13.

Asian shares rose on Monday, anticipating a week of earnings reports and central bank meetings that could lead to easing in the US and UK, while Japan might increase borrowing costs.

SpaceX will move Dragon spacecraft splashdowns to the West Coast in 2025 to better control reentering debris from the trunk section.

Galaxsys has received official licensing in Sweden, marking a key milestone in its expansion into regulated markets and commitment to high-quality, responsible gaming.

Don’t Miss - The Russo brothers will direct two new Avengers films, “Avengers: Doomsday” and “Avengers: Secret Wars,” as announced at San Diego Comic-Con. “Doomsday”, featuring Robert Downey Jr. as Doctor Doom, is set for release in May 2026, followed by “Secret Wars” in May 2027

Submit your artwork to Mindstream →

Just a good ol' PR move. - 52% ✅

This will set new industry standards! - 48%

Your Views:

“Apple's commitment to safe and secure AI development with the White House is a smart move. Integrating generative AI into its products while adhering to guidelines ensures user trust and regulatory compliance. This step signifies Apple's dedication to leading in responsible AI innovation.” - Tjord

“I don't see Apple's AI commitments as groundbreaking. While it’s good they’re aligning with the White House guidelines, it feels more like a PR move than a substantial change. The integration of generative AI into iOS is interesting, but not revolutionary.” - alex

Submit your opinions in our daily poll to be featured!