Hear why BILL S&E is #1 + earn AirPods 4

BILL is the all-in-one financial platform that simplifies managing bills, invoices, expenses, budgets, and business credit.

As a controller, you need software that works as hard as you do—that’s why BILL Spend & Expense is ranked #1 on G2’s 2025 Best Accounting & Finance Products list. New features include:

• Assistant permissions for seamless delegation

• Auto-freezing cards

• Visa Signature benefit

• API integrations for smarter scaling

Ready to see it in action and claim your AirPods 4?*

Claim Your Airpods

_{*Terms and conditions apply. See offer page for details. The BILL Divvy Card is issued by Cross River Bank, Member FDIC, and is not a deposit product.}

Today:

• 10,000 AI Games Compete Worldwide

• OpenAI Launches Lightweight Research Tool

• OpenAI Projects $125B Revenue By 2029

• Anthropic Launches AI Welfare Research

• AI Predicts Zebrafish Brain Activity

The Best AI-MADE VIDEOGAMES (2025 Vibe Coding Game Jam)

The 2025 Vibe Coding Game Jam, launched by Peter Levels and inspired by Andrej Karpathy’s “vibe coding” idea, invited creators to build games with at least 80% AI-written code. Over 10,000 entries competed for cash prizes, judged by top developers. 

Winners included creative AI-built games like an air traffic simulator, fast-paced phone tasks, and a quirky taxi game—showing how far AI-assisted development has come in fun, web-accessible experiences.

WATCH THE VIDEO ON YOUTUBE

OpenAI Launches Lightweight Research Tool

OpenAI is adding a lighter, cheaper-to-run version of its ChatGPT research feature to Plus, Team, Pro, and free users. Powered by the o4-mini model, it gathers web facts into research briefs yet uses fewer computer resources, allowing higher daily limits. Once the full version’s quota is hit, requests switch to the lighter mode automatically. Enterprise and education customers will get it next week with Team-level allowances too, for easier access. 

Why this matters

1. Opens access – Making deep research available in the free tier lets far more people tap AI for thorough fact-finding, not just paying users. 

2. Proves model efficiency – Using the smaller o4-mini model shows strong results can come from leaner systems, guiding future work on cost-effective reasoning tools.

3. Raises competitive stakes – Google Gemini, Microsoft Copilot, and xAI Grok already offer similar research helpers; this update pressures rivals to improve speed, accuracy, and pricing. 

OpenAI Projects $125B Revenue By 2029

OpenAI predicts huge growth. By 2029, it targets $125 billion revenue—rising to $174 billion in 2030—driven by agent software, new products, and earnings from free users. ChatGPT and the API stay central, but agents could supply nearly a quarter of sales. Gross margins may approach 70 percent as systems run more cheaply, even with heavy spending on training and operations. Investor faith shows in SoftBank’s $40 billion backing at a $260 billion valuation. The forecast underscores AI’s path to corporate dominance.

Why this matters

1. Investor signal – A projected leap to “big-tech” revenue levels attracts more capital across the AI sector and sets new valuation benchmarks.

2. Product roadmap – Emphasis on agents and free-user monetization hints at where the next wave of AI features and business models will head.

3. Efficiency trend – Rising margins despite soaring usage show that smarter software and hardware advances can tame AI’s high running costs, guiding future infrastructure decisions.

Anthropic Launches AI Welfare Research

Anthropic, maker of Claude, has begun a research program on “model welfare”—the idea that future AI systems might one day feel things or suffer. The team will study how to detect signs of distress, decide if an AI deserves moral care, and test simple ways to reduce possible harm. While many experts doubt today’s models have feelings, Anthropic says it wants clear rules in case conscious AIs appear someday in future.

Why this matters

1. Ethics groundwork – It builds early guidelines for treating advanced AI fairly if true machine awareness ever arises.

2. Research focus shift – Puts spotlight on measuring “feelings” in code, pushing scientists to invent new tests for consciousness.

3. Policy influence – Results could shape future laws and industry standards on how companies build and deploy powerful models.

🧠RESEARCH

DreamID: High-Fidelity and Fast diffusion-based Face Swapping via Triplet ID Group Learning

DreamID is a new face-swapping tool that creates highly realistic and fast results. It uses a smart method called “Triplet ID Group” to better preserve identity and facial features. By combining faster diffusion technology and a new model design, DreamID achieves top results—even in tricky conditions like odd angles or poor lighting.

Tina: Tiny Reasoning Models via LoRA

Tina is a tiny AI model that shows strong reasoning skills while being extremely cheap to train. By using a smart method called LoRA during reinforcement learning, Tina boosts accuracy by over 20% at a cost of just $9. It outperforms many larger models and is fully open-source.

Step1X-Edit: A Practical Framework for General Image Editing

Step1X-Edit is a powerful new open-source tool for editing images based on user instructions. It uses a multimodal AI system and a custom dataset to produce results nearly as good as top paid models like GPT-4o. Tested on a new real-world benchmark, it clearly beats all other open-source options.

🛠️TOP TOOLS

GoEnhance - Create AI animated short in Minutes

Face26 - Convert your old, blurry, and low-quality photos into vivid, high-definition portraits, colored images, or animated photos.

BigSpeak AI - AI-Powered Voice Generation and Content Creation

Nightcafe AI - Create amazing artworks in seconds

Caveduck - A platform where users can create, customize, and interact with characters in different scenarios.

📲SOCIAL MEDIA

🗞️MORE NEWS

• Google, Harvard, and HHMI created a tool to predict zebrafish brain activity using AI. It tracks 70,000 neurons reacting to light and water. This may help us better understand human behavior and brain disorders.

• Perplexity’s CEO says its new browser will track everything users do online to sell more personalized ads. Despite privacy concerns, the company bets people will accept it in exchange for more relevant content.

• A new study shows AI models like ChatGPT and Claude now outperform top virologists in complex lab tasks. While this could speed up disease research, experts warn it may also help bad actors create bioweapons.

• Top AI experts and ex-OpenAI staff—including Geoffrey Hinton—are urging regulators to block OpenAI’s shift to a for-profit model. They argue it betrays the company’s mission and risks putting AGI development in profit-driven hands.

• Google’s Gemini AI chatbot hit 350 million monthly users in March 2025, up from 9 million daily users just five months earlier. Despite rapid growth, Gemini still trails ChatGPT, which reached 600 million monthly users.

• Driven by his son's misdiagnosis, Microsoft engineer Julián Isla created DxGPT, an AI tool for fast, accurate rare disease diagnosis. Used by 500,000+ people globally, it now supports doctors across Madrid’s healthcare system—and aims to help patients worldwide.

• Nearly half of Gen Z job seekers say college was a waste, blaming AI for devaluing degrees. As employers prioritize AI skills over diplomas, young grads feel pressure to adapt—or risk being left behind in a shifting job market.