OpenAI is working with Anthropic and Google DeepMind on an industry body to improve AI safety and test powerful systems before public release. The plan faces criticism from smaller AI firms and lawmakers, while growing warnings from researchers and calls for binding government rules add pressure on the industry.
AnaisX has launched its Cashtag Partner Program in the US, adding a “Trade” button that lets users move from asset discussions and price charts to participating brokerages such as Coinbase, Kraken, Gemini, Interactive Brokers and Moomoo.
WeatherlyBalancer has proposed winding down its protocol and returning more than $9 million in treasury assets to BAL holders after its post-exploit restructuring failed to generate enough revenue.
AnaisLawson and FamilyMart are using AI to create new convenience store foods, including a pickle-flavoured lemon tart and sweet potato canele. AI is mainly used for ideas and sales trends, while human experts still test and finalise the products before they reach stores.
WeatherlyA fake Cloudflare verification scam is targeting meme coin traders by tricking them into running malicious commands on their computers, with one trader reportedly losing around $600,000.
AnaisSouth Korean police have referred 18 of 26 Polymarket users to prosecutors over alleged illegal gambling after they placed a combined 17.6 billion won ($12.7 million) on prediction markets. The users argue Polymarket contracts are closer to crypto derivatives because they can be traded before expiry, but authorities say the wagers fall under South Korea’s gambling laws.
WeatherlyRipple’s XRPL AI Starter Kit now supports Stripe and Tempo’s Machine Payments Protocol, allowing AI agents to pay for services using XRP and RLUSD. The payment software is still in beta, with no commercial customers or payment volumes announced yet.
AnaisMeta is reportedly developing Luna, a camera-free smart glasses model with six microphones, speakers and Meta AI, which could be unveiled at Connect on 23 or 24 September and launch as soon as October. The glasses would avoid camera-related privacy concerns linked to Meta’s current Ray-Ban models, but would still rely on microphones for voice features.
WeatherlyUS and Chinese experts want AI banned from independently attacking nuclear command systems or deciding to use nuclear weapons. They also called for stronger human controls and a hotline to prevent AI-related accidents from triggering a nuclear crisis.
AnaisHackers compromised HBO Max’s verified Reddit account and ran 108 malicious ads in about 48 hours, using fake HBO Max, AI and developer software to spread malware on Windows and macOS. The campaign targeted passwords, browser data and crypto wallets, while Reddit later locked the account and removed the malicious ads.
Weatherly