|
|
Hi there, this is your daily ☕️ Techpresso.
|
|
|
In today's Techpresso:
|
|
💥 OpenAI launches ChatGPT Translate to rival Google Translate 🚫 X blocks Grok from generating sexualized images of real people 🔄 Two Thinking Machines Lab cofounders return to OpenAI 📚 Microsoft and Meta pay Wikipedia for AI training data 📰 Digg relaunches as new Reddit competitor 💰 OpenAI signs $10 billion computing deal with Cerebras 🎁 + 18 other news & articles you might like 🧰 + 6 trending tools 📚 + 5 trending papers & reports
|
|
FROM OUR PARTNER
As AI adoption grows, new threats target AI systems across hosting environments, supply chains, and interfaces. Datadog's AI Security Best Practices Guide breaks down how to secure: - The underlying infrastructure that hosts and runs AI applications
- The software and data layers AI applications rely on to operate
- The entry points and business logic that enable users and systems to interact with AI
Build a more resilient AI environment with clear, practical security guidance you can apply today. Read the eBook to learn more
|
💥 OpenAI launches ChatGPT Translate to rival Google Translate
LINK
|
- OpenAI has released ChatGPT Translate, a new standalone translation tool that works like Google Translate but adds AI-powered features to help users adjust their translated text for different purposes.
- The tool offers one-tap prompt options that let users reshape translations to sound more fluent, formal, simple for children, or suited for academic readers, then opens ChatGPT for deeper changes.
- ChatGPT Translate currently lacks key features Google offers, including document uploads, website translation, real-time conversations, and broad language support, with Google backing over 50 more languages total.
|
🚫 X blocks Grok from generating sexualized images of real people
LINK
|
- X has blocked Grok from generating sexualized images of real people after users exploited the AI chatbot to create non-consensual explicit pictures, including images of minors, often by tagging Grok directly under photos posted on the platform.
- The company now restricts image creation and editing through Grok to paid subscribers only, and has added geoblocking in places where generating images of real people in bikinis or underwear is illegal.
- Despite these changes, Grok can still remove or alter clothing from uploaded photos, and regulators in California, the UK, Australia, and several other countries have opened investigations into xAI over potential law violations.
|
🔄 Two Thinking Machines Lab cofounders return to OpenAI
LINK
|
- Two cofounders of Thinking Machines Lab, the AI startup led by former OpenAI CTO Mira Murati, are returning to OpenAI, along with a third key employee named Sam Schoenholz.
- Thinking Machines Lab fired cofounder Barret Zoph before his departure, citing "unethical conduct" and sharing "confidential company information with competitors," though OpenAI CEO Fidji Simo says she has no such concerns.
- The startup has now lost half its cofounders while reportedly trying to raise funds at a $50 billion to $60 billion valuation, which could hurt both investor interest and future recruiting efforts.
|
📚 Microsoft and Meta pay Wikipedia for AI training data
LINK
|
- Microsoft, Meta, Amazon, and AI startups Perplexity and Mistral AI are now paying Wikipedia through its enterprise product to access content for training their AI models.
- Tech companies scraping Wikipedia's 65 million articles for free has increased server demand and costs for the non-profit, which relies mainly on small public donations to operate.
- The Wikimedia Foundation spent time developing the right features to move companies from free access to a paid commercial platform that handles their large-scale AI training needs.
|
📰 Digg relaunches as new Reddit competitor
LINK
|
- Digg, the early internet community that once competed with Reddit, is relaunching as a public beta on Wednesday under the ownership of its original founder Kevin Rose and Reddit co-founder Alexis Ohanian.
- The new platform will use zero-knowledge proofs and signals from mobile devices to verify real users and build trust, aiming to address social media toxicity and prevent AI bots from taking over.
- Anyone can now join and create their own communities on any topic, with community managers able to set rules and share moderation logs publicly so members can see what decisions are made.
|
💰 OpenAI signs $10 billion computing deal with Cerebras
LINK
|
- OpenAI has signed a multi-year deal worth over $10 billion with AI chipmaker Cerebras, which will provide 750 megawatts of compute power to the company starting this year through 2028.
- Both companies say the partnership will deliver faster outputs for OpenAI's customers, with Cerebras adding a dedicated low-latency inference solution that speeds up responses requiring more processing time.
- Cerebras, which claims its AI-specific chips are faster than GPU-based systems like Nvidia's, has delayed its 2024 IPO while reportedly seeking another billion dollars at a $22 billion valuation.
|
|
FROM OUR PARTNER
Medeo is the world's first conversational AI video creation tool Describe your idea, and Medeo's AI Editing builds production-ready files with full timeline control—generating scripts, structuring scenes, matching visuals, voice and music. Adjust any shot, pace, or style by natural language. No editing skills are needed. Maintain character consistency across all scenes and support video creation across all categories. Try Medeo for free. |
|
|
|
Other
news & articles you might like
-
Apple will pay billions for Gemini; OpenAI decided against Siri deal – FT
LINK
-
2026 May Be the Year of the Mega I.P.O.
LINK
-
Oracle sued by bondholders over losses tied to AI buildout
LINK
-
Amazon launches its 'sovereign' cloud in Europe and plots expansion
LINK
-
Google’s Trends Explore page gets new Gemini capabilities
LINK
-
App downloads declined again in 2025, but consumer spending soared to nearly $156B
LINK
-
McKinsey tests AI chatbot in early stages of graduate recruitment
LINK
-
Google plans to make Chrome for Android an agentic browser with Gemini
LINK
-
Amid the AI boom, Intel is focusing on the basics
LINK
-
Despite OpenAI partnership, Microsoft is one of Anthropic's biggest customers
LINK
-
Anthropic cuts AI productivity forecasts in half after analyzing Claude's real-world failure rates
LINK
-
A single click mounted a covert, multistage attack against Copilot
LINK
-
Can Spotify strike the right chord with AI?
LINK
-
Claude Cowork Exfiltrates Files
LINK
-
Why no one can agree on what AI will do to jobs
LINK
|
|
🧰 Trending tools
|
TeamOut AI: an AI copilot that streamlines corporate offsite planning by providing instant venue recommendations, flight estimates, and hotel deals with up to 40% savings.
LINK
|
|
Cal.com Companion Apps: native mobile apps and browser extensions that let you manage scheduling and calendar bookings directly from iOS, Android, or your browser toolbar.
LINK
|
|
AI Test Engineer by BlinqIO: automates end-to-end test creation, maintenance, execution, and analysis using GenAI to eliminate testing bottlenecks in software release cycles.
LINK
|
|
Binary: a lightweight applicant tracking system that streamlines hiring with video questions, AI-powered candidate reviews, and custom job pages for startups.
LINK
|
|
LyzrGPT: a private enterprise AI chat platform that runs in your own infrastructure, letting you switch between AI models while keeping data secure and internal.
LINK
|
|
Grsh: a minimal, memory-safe command-line shell written in Rust, designed primarily for FreeBSD with macOS compatibility.
LINK
|
|
You can check the previous tools here, or add your tool here
|
|
📚 Trending papers & reports
|
Techpresso is hiring its first Account Executive: If you've got energy, drive, and want to join a fast-growing tech newsletter — we want to hear from you. Apply Here
|
|
Prompt injection attacks now spread like viruses: researchers discovered malicious prompts can self-replicate across AI systems, creating the first documented case of AI-based malware behavior.
LINK
|
|
Interstellar visitor shattered before reaching sun: researchers found the third known object from outside our solar system broke apart into dust before its close approach.
LINK
|
|
Quantum computers could verify math proofs faster: researchers show quantum algorithms can check certain mathematical proofs exponentially quicker than classical computers for specific problem types.
LINK
|
|
Cascade distillation creates smaller AI models: researchers compressed a 123-billion parameter model down to 3 billion while maintaining 96% of original performance through sequential pruning.
LINK
|
|
Mathematicians calculate algebraic structure of geometric maps: the team computed a fundamental mathematical invariant describing how spheres can wrap around complex flag varieties in algebraic geometry.
LINK
|
|
|
|
|
