AI & Web3 Weekly: Nvidia’s Record Surge, Political AI PACs, and Web3 Security Shocks

This week, Nvidia defies export restrictions with a 56% Q2 revenue jump, OpenAI pilots in-chat shopping, and WeTransfer backtracks on AI training policies. In Web3, JPMorgan bets $500M on Numerai, researchers warn of AI agent vulnerabilities, and new exploits rock DeFi platforms.

Author

Azhar Uddin
September 01, 2025

AI Snippets

Nvidia Q2 revenue jumps 56% despite export restrictions

Nvidia defied U.S.–China export controls with a stunning 56% revenue increase in Q2, underscoring overwhelming demand for AI chips even amid geopolitical friction.

READ MORE →

AI‑focused super PAC emerges ahead of 2026 elections

Backed by investors and firms like Andreessen Horowitz and Perplexity, a new “Leading The Future” super PAC is pouring funding into pro‑AI political candidates—marking a high‑stakes fusion of AI and electoral politics.

OpenAI pilots in‑ChatGPT checkout that could reshape e‑commerce

OpenAI is testing a shopping feature directly inside ChatGPT that lets users buy products without leaving the interface—potentially altering how brands surface online and monetizing AI search.

WeTransfer reverses user‑content AI training policy amid backlash

After facing pressure from creatives, WeTransfer backed off plans to use user‑generated content to train AI—highlighting how content rights and online creativity remain central to AI ethics debates.

Web3 Snippets

JPMorgan backs AI‑powered hedge fund Numerai with $500M

Numerai, the hedge fund combining decentralized data with AI models, secured a $500 million investment from JPMorgan—signaling big‑bank confidence in AI‑infused crypto strategies.

Scalable AI‑Web3 convergence: study maps 133 projects

A fresh academic survey classifies 133 projects at the intersection of AI agents and Web3, exploring applications in DeFi optimization, security auditing, governance, and trust frameworks—pointing to a maturing, multidimensional ecosystem.

Researchers expose AI agent vulnerabilities in decentralized finance

A security analysis found that AI agents in blockchain systems can be manipulated through subtle prompt or memory tampering, triggering unintended asset transfers—underscoring the need for rigorous defenses.

Sentinel Labs: AI‑generated bot scams drain $1M from users

Malicious actors are deploying AI‑crafted YouTube videos promoting “free MEV bots,” which actually siphon funds through hidden code—highlighting an urgent cyber‑security threat in Web3.

Credix DeFi protocol disappears after $4.5M exploit

Credix suffered a $4.5 million exploit when a hacker—or possibly insider—drained liquidity. The team pledged repayments, then vanished entirely, igniting suspicions of a rug‑pull.

Abra pauses international withdrawals amid risk concerns

Crypto lender Abra abruptly halted its international Earn services, citing risk management—and reviving alarm bells from earlier pauses and regulatory action.