|
|
Hi there, this is your daily ☕️ Techpresso.
|
|
|
In today's Techpresso:
|
|
❌ Google says Gemini won’t have ads 📄 Internal documents expose Microsoft OpenAI partnership secrets 👀 OpenAI rolls out age prediction on ChatGPT 🎬 YouTube CEO makes fighting AI slop a top priority for 2026 🔓 X open sources its algorithm 🧠 Meta AI lab delivers first major models internally 🎁 + 16 other news & articles you might like 🧰 + 6 trending tools 📚 + 5 trending papers & reports
|
|
FROM OUR PARTNER
IBM explores how enterprises can pivot from incremental gains to exponential growth. Discover why innovation—not optimization—will define winners in 2030 and how custom AI models will create lasting differentiation. - Reinvent fast: Disrupt markets quarter by quarter.
- Custom AI portfolios: Outperform generic models.
- Future-proof strategy: Quantum and AI converge.
Read the Enterprise 2030 Report
|
❌ Google says Gemini won’t have ads
LINK
|
- Google DeepMind CEO Demis Hassabis confirmed at Davos that the company has no plans to add ads to its Gemini chatbot anytime soon, just weeks after OpenAI started testing ads in ChatGPT.
- Hassabis suggested OpenAI may need extra revenue, while Google can keep Gemini ad-free because it already earns over $200 billion yearly from advertising through its search business and AI Overviews.
- Hassabis also confirmed that Gemini models will soon power the next generation of Siri through a partnership with Apple, a deal that helped Alphabet's valuation pass Apple's for the first time since 2019.
|
📄 Internal documents expose Microsoft OpenAI partnership secrets
LINK
|
- Internal documents from Elon Musk's lawsuit against OpenAI reveal how Microsoft pursued and invested in the AI lab over the past decade, showing private emails and messages between top executives.
- Microsoft initially lost OpenAI to Amazon Web Services in 2015, but won them over in 2016 with a $50 million compute donation, and later invested $1 billion in 2019 after OpenAI created its capped-profit structure.
- During the November 2023 board crisis that fired Sam Altman, text messages show Microsoft executives helped screen potential board members in real time, despite having no official seat on the board.
|
👀 OpenAI rolls out age prediction on ChatGPT
LINK
|
- OpenAI has launched its "age prediction" feature for ChatGPT users around the world, aiming to spot minors and automatically apply content restrictions to their conversations with the chatbot.
- The feature uses an AI algorithm that checks "behavioral and account-level signals" like stated age, how long an account has existed, and what times of day someone typically uses ChatGPT.
- If the system wrongly flags an adult as underage, they can fix it by submitting a selfie to OpenAI's ID verification partner Persona, which will restore full access to their account.
|
🎬 YouTube CEO makes fighting AI slop a top priority for 2026
LINK
|
- YouTube CEO Neal Mohan announced in his annual letter that fighting low-quality "AI slop" content will be a major focus for 2026, while also expanding AI tools for creators.
- The platform plans to build on existing systems that combat spam and clickbait to reduce the spread of low-quality, repetitive AI-generated content across YouTube.
- Mohan also highlighted new AI features coming this year, including tools letting creators make Shorts using their likeness and autodubbing, plus stronger parental controls for kids.
|
🔓 X open sources its algorithm
LINK
|
- X has open sourced its recommendation algorithm for the second time, sharing code and documentation on GitHub that explains how the platform decides which posts appear in user feeds.
- The system relies entirely on X's Grok-based transformer to learn what users find relevant by tracking their engagement history, with no manual human adjustments to how content relevance is determined.
- Critics previously called the 2023 code release "transparency theater," and X faces ongoing scrutiny including a $140 million EU fine for violating transparency rules and pressure over Grok-generated explicit content.
|
🧠 Meta AI lab delivers first major models internally
LINK
|
- Meta's new AI research group, Meta Superintelligence Labs, has delivered its first important models to internal teams this month, according to CTO Andrew Bosworth speaking at the World Economic Forum in Davos.
- The team has been working for about six months, and Bosworth called the AI models "very good," though he did not say which specific projects were completed or mention the codenames Avocado or Mango.
- Meta had faced criticism over its Llama 4 model's performance, and Bosworth noted the technology still needs significant post-training work before it can be released to consumers or used internally.
|
|
FROM OUR PARTNER
The Techpresso team tested Circleback for weeks, and it's the best AI note-taking tool we've used. Unlike competitors, it doesn't just transcribe. It extracts what matters: action items automatically organized so you know exactly what to do next. The game changer? Circleback integrates with your tools and auto-updates your CRM after calls. The biggest win? We stopped scribbling notes and started listening. Sign up with your email for one month free |
|
|
|
Other
news & articles you might like
-
OpenAI starts selling ChatGPT ads, charges by views instead of clicks
LINK
-
Elon Musk says Tesla’s restarted Dojo3 will be for ‘space-based AI compute’
LINK
-
The FTC isn't giving up on its antitrust case against Meta
LINK
-
Palantir CEO says AI ‘will destroy’ humanities jobs but there will be ‘more than enough jobs’ for people with vocational training
LINK
-
Gemini AI assistant tricked into leaking Google Calendar data
LINK
-
Snap reaches settlement in social media addiction lawsuit
LINK
-
Adobe Acrobat now lets you edit files using prompts, generate podcast summaries
LINK
-
Amazon CEO Andy Jassy goes wobbly on AI bubble possibility
LINK
-
Burritos from Heaven: Are drones the future of delivery?
LINK
-
Salesforce's Benioff calls for AI regulation after recent suicides
LINK
-
Consumers spent more on mobile apps than games in 2025, driven by AI app adoption
LINK
-
Anthropic study finds that role prompts can push AI chatbots out of their trained helper identity
LINK
-
Wikipedia volunteers spent years cataloging AI tells. Now there's a plugin to avoid them.
LINK
-
The World's First Commercial Space Station Is Nearly Ready to Launch
LINK
|
|
🧰 Trending tools
|
Blink Agent Builder: a platform that generates AI agents from natural language descriptions, with built-in tools like web search, code execution, and 180+ models for autonomous task completion.
LINK
|
|
Mastra: an open-source TypeScript framework for building AI applications and agents with integrated workflows, memory management, and tool orchestration capabilities.
LINK
|
|
Citable: tracks how AI platforms like ChatGPT and Perplexity reference your brand, helping optimize visibility in AI-generated responses and search results.
LINK
|
|
FiloMail for Android: a Gmail client that auto-summarizes threads, suggests replies, creates tasks with due dates, and uses AI-powered filters to automatically organize your inbox.
LINK
|
|
Modo: generates hardware project designs from ideas by matching real components and auto-sizing enclosures using actual datasheet dimensions to speed up prototyping.
LINK
|
|
Agent Skills Leaderboard: a benchmarking platform that ranks AI coding agents based on their performance across standardized programming tasks and challenges.
LINK
|
|
You can check the previous tools here, or add your tool here
|
|
📚 Trending papers & reports
|
The habits of high-performing mobile teams: The Bitrise Mobile Insights Report breaks it down using real customer data, showing what drives app development success and what sets high-performing teams apart. Get the full report.
|
|
Private likes on X increased risky engagement: after X made likes private, users were 2.4 times more likely to like posts about taboo topics they previously avoided publicly.
LINK
|
|
Detecting Malicious Software Through Execution Patterns: the system mathematically proves whether programs are malicious by watching how they run, catching threats that hide their code until execution.
LINK
|
|
ChatGPT users form emotional bonds despite knowing it's AI: people share secrets, seek comfort, and feel genuine connection even while acknowledging the system has no real feelings.
LINK
|
|
DiffRatio Generates Images in One Step: diffusion model creates high quality images without the usual multi-step process, cutting memory usage in half by predicting ratios instead of noise.
LINK
|
|
Generating 3D models from single photos: the system creates consistent 3D objects by learning how lighting and materials affect appearance across different viewing angles.
LINK
|
|
|
|
|
